Benchmark Almanac

JevBench

GPT-6 Luna 98.4% leads since September 2026
Not in model cards

Decision models, such as TypeSafe AI's Jev and compatible systems, answer typed decision questions. The board scores accuracy, probability calibration, latency and cost.

Maintained independently of TypeSafe AI. Show HN post had 154 points. Related to the Decision Index board.

Released
September 2026
Kind
Other
Level
Everyday
Scored
Checked automatically
Answers
Prediction
Tasks
All public
Strongest model tested
DeepSeek V4.1 Flash #34 on the capability index
Last model added
September 2026
Scores refreshed
November 12th, 2026
339
Entries on its own leaderboard, 4 models we track

GPT-6 Luna leads with 98.4% among models released by September 2026

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

339 entries. Scroll sideways for the rest.

  1. 98.4 GPT-6 Luna
  2. 93.6 GPT-5.6 Luna
  3. 91.8 Torchcast Decision 31B v1.0
  4. 90.7 MetaDecision-31B-Rain
  5. 89.6 DeepSeek V4.1 Flash
  6. 89.5 Karpo-Decision-31B
  7. 88.7 WeMM-JEV (Rover00)
  8. 88.2 Jeff-1.1-Large (jgeuter)
  9. 86.5 VisionJEV 1.0.0
  10. 86.3 kettei-fat (Sqwish)
  11. 86.0 Piccolo 27B (agentscope-ai)
  12. 85.8 Super Decider (thinking)
  13. 85.2 baselabs mammoth-1 (27B)
  14. 85.2 KnowLine 31B-Gen1
  15. 84.9 Fastino GLiDE 1.2 (open weights, self-hosted)
  16. 84.9 Super Decider
  17. 84.5 Fastino GLiDE (hosted API, 22 Oct 2026)
  18. 84.3 H2O-Lightning-31B v1.0.7 (H2O.ai, Gemma-4-31B-it fine-tune, vLLM + h2oai overlay, one-engine FP8 profile, self-hosted)
  19. 84.2 Drex Max (Nace AI)
  20. 84.0 decide-1.1 (decide.sh, API)
  21. 83.9 oenogarum (plantwaterer; commercially restricted)
  22. 83.2 allec-oavg (plantwaterer; commercially restricted)
  23. 82.7 Perplexity Decider v1.1 27B (Qwen3.8-27B, noncausal decision readout)
  24. 82.6 vLLM-SR d3-turbo
  25. 82.6 StartLux-Decision v1.1 27B Instant (StartLux Labs, Qwen3.8-27B fine-tune)
  26. 82.3 Decision-King v1.1 (xsdfasfgsa)
  27. 82.2 vLLM-SR d3
  28. 82.1 basal-omni-1.0-ultra (Remek Kinas, Gemma-4-31B-it + merged LoRA, typed-decision readout, self-hosted)
  29. 82.0 DOPEST 1.0 31B
  30. 81.9 SubQ Snap 1.1 (SubQ)
  31. 81.9 Jeff-1.1-Medium (jgeuter)
  32. 81.8 Arcbright Elf 1 31B
  33. 81.7 Quyet-1.0-Large (Chinh Nguyen, Gemma-4-31B decoder, option-letter logits)
  34. 81.7 Jeffan-31b-1710
  35. 81.5 Jeff-1.0-Large
  36. 81.5 SubQ Snap (preview, SubQ)
  37. 81.4 AJev lm6q (Gemma 4 31B)
  38. 81.2 Nimble-v4-max (open weights, self-hosted)
  39. 81.1 Sieve-31B-v2
  40. 80.9 Turion Jev 31B
  41. 80.8 pplx-decider-v1.2 (Perplexity, Qwen3.8-27B full fine-tune, marker readout, self-hosted)
  42. 80.5 d1-ultra (Liquid AI)
  43. 80.2 torchcast-decision-27b (Torchcast AI, Qwen3.8-27B fine-tune)
  44. 79.9 Arcbright Elf 1 12B (r2)
  45. 79.8 Lychee-27B
  46. 79.8 Microsoft-Decision-1 (updated weights, Oct 2026)
  47. 79.7 Bespoke Nimble-v4-max (Bespoke Labs)
  48. 79.6 Jeff-1.1-Qwen-4B (jgeuter)
  49. 79.6 metask-jev-rain-31B-trained service v3
  50. 79.6 decisio v0.8.0 on gemma-4-31B-it (frozen, FP8 on load, one prefill per state, self-hosted)
  51. 79.5 Vera Large
  52. 78.9 Kev 27B (Kev 1.0, Qwen3.8-27B)
  53. 78.9 metask-jev-rain-31B-trained
  54. 78.8 Myles-G 31B (Launchables)
  55. 78.8 StartLux-Decision v1.1 27B AutoThink (StartLux Labs, Qwen3.8-27B fine-tune, gate 0.6)
  56. 78.7 Decisio 12B tuned commit-think (variant of the Decisio family)
  57. 78.6 Sage 1.3.0 (Levanto Labs)
  58. 78.5 SPX-CD Pro (SurdAI, hosted /v1/systemone, Oct-4 checkpoint)
  59. 78.3 metask-jev-rain-12B
  60. 78.3 classifier.dev (fast tier)
  61. 78.0 Amos 1 large (Scira)
  62. 77.9 Bobcat 1.3 (Gemma 4 31B)
  63. 77.6 deck-31B (krishna765, frozen Gemma-4-31B-it, TorchAO FP8 dynamic)
  64. 77.5 SD-JEV-4B-v12.1.0 (Kego, author-hosted API)
  65. 77.5 Torchcast Decision 4B (v1.3)
  66. 77.3 KnowLine 26B-A4B-Gen2.5
  67. 77.2 Eikos-27B (caiovicentino1, Qwen3.8-27B)
  68. 77.1 Jev 1.13.0 (TypeSafe AI)
  69. 77.1 wity-1 (Wity, reasoning always)
  70. 77.0 H2O-Lightning-4B v1.4.1
  71. 77.0 Fastino GLiDE
  72. 76.9 wity-1 (Wity, reasoning auto)
  73. 76.9 metask-jev-rain-12B-trained
  74. 76.9 Aleria Judge 1.1.1 (ALERIA-ML, inference technique on stock Gemma 4 31B, author server, self-hosted)
  75. 76.8 Wald 26B-A4B (Gemma 4 26B-A4B)
  76. 76.8 DecisionNode
  77. 76.7 KnowLine 35B-A3B-Gen1.5
  78. 76.6 Piccolo 9B (agentscope-ai)
  79. 76.5 cribl-decision-1.0-27B (Cribl)
  80. 76.0 René-1 31B FP8 (salfatigroup, Gemma 4 31B full fine-tune, one-pass option readout)
  81. 75.8 OpenJev (thinking, BF16)
  82. 75.7 Decisio 12B tuned commit (variant of the Decisio family)
  83. 75.3 Clef (Cloudflare, Qwen3.8-27B post-train with a joint schema head, multimodal, measured on text)
  84. 75.0 Jev 1.14.0 (TypeSafe AI)
  85. 75.0 H2O-Lightning-4B v1.1 (H2O.ai, Qwen3.5-4B fine-tune, stock vLLM + open shim)
  86. 74.9 SimpleJev Qwen3.8-27B (self-hosted, v1.6 pool)
  87. 74.5 Umpire 35B-A3B
  88. 74.5 z3-mini-soup-preview (Zils AI)
  89. 74.4 d1 (Liquid AI)
  90. 74.3 SPX-CD Flash (SurdAI, hosted /v1/systemone, Oct-4 checkpoint)
  91. 74.0 iota-decide-4b-v0.4 (Macrocosmos, letter-readout 4B, self-hosted)
  92. 73.9 JEV-27B (AutoTrust, Qwen3.8-27B)
  93. 73.8 Nagi-ENORMOUS 27B (nagisanzenin, Qwen3.8-27B LoRA)
  94. 73.6 Mercury Decide (Inception; System One decisions API, served free on OpenRouter as inception/mercury-decide:free)
  95. 73.6 Celeris-1 Decision (Celeris)
  96. 73.6 Piccolo 4B (agentscope-ai)
  97. 73.5 OpenAI Decisions (gpt-6-luna)
  98. 73.5 Surogate Rune 26B-A4B v3 (v1.6 pool)
  99. 73.5 RSI-Jev v6.1-VL 27B
  100. 73.3 Sage 1.4 (Levanto Labs)
  101. 73.1 kettei-chub (Sqwish)
  102. 73.0 Bobcat Flash 1.2 (Gemma-4-26B-A4B-it + two merged rank-64 LoRA adapters, typed-decision readout at the first answer position, FP8, self-hosted)
  103. 72.7 Autoloops – Gemma 4 31B IT
  104. 72.6 Torchcast Decision 4B (v1.2)
  105. 72.4 swanOne (blockbrain, Qwen3.8-Flash-Next NVFP4)
  106. 72.3 FrontiersMind Lumma-fev-27B
  107. 72.0 Microsoft-Decision-1 (Azure Foundry) — A2 re-run
  108. 71.7 torchcast-decision-12b (Torchcast AI, Gemma-4-12B fine-tune, option-letter logprob readout)
  109. 71.6 AutoJev-27B (denis-pplx, Qwen3.8-27B)
  110. 71.6 SimpleJev Qwen3.8-27B
  111. 71.3 decider-12b v2 (Mapika)
  112. 71.3 Decision 2.0 Vega 27B (vLLM-SR, Qwen3.8-27B)
  113. 71.3 Jev-Omni (akhilaaa3, Gemma-4-12B merged)
  114. 71.3 Winnow-12B Q8
  115. 71.2 Xor 26B-A4B (Juspay, Gemma-4-26B-A4B, bf16)
  116. 71.0 decider-12b v1, stock Gemma-4-12B-it (Mapika)
  117. 70.9 Cygnet (blockbrain, frozen Gemma-4-12B-it)
  118. 70.8 Microsoft-Decision-1 (Azure Foundry)
  119. 70.7 decisio v0.8.0 on gemma-4-12B-it (frozen, one forward pass, self-hosted)
  120. 70.7 JPT-35B-A3B (Qwen3.5-35B-A3B fine-tune)
  121. 70.7 Myles-G 12B (Launchables)
  122. 70.5 Instinct (ZooWork, Qwen3.8-27B)
  123. 70.5 decider chat on Gemma-4-31B-it (Mapika, frozen base, inference technique)
  124. 70.3 baselabs elephant-1 (9B)
  125. 70.2 Jebadiah 27B (Frontier Infra, Qwen3.8-27B LoRA)
  126. 70.1 GEV-26B-Decide (AutoTrust, Gemma-4-26B-A4B + LoRA + head)
  127. 70.1 Chon-1.0-12B-Fast
  128. 70.0 Blink v0.3 26B-A4B
  129. 70.0 decisio v0.9.0 on gemma-4-12B-it (frozen, one prefill per question, self-hosted)
  130. 69.7 NInfer Qwen3.8-Flash-Next mixed
  131. 69.5 attest-31B-v1.0
  132. 69.3 Respan Span-02
  133. 69.3 Nemotron Deciffusion 8B v2 (pst2154)
  134. 69.3 Philly Decider v2 (PhillyLabs)
  135. 69.3 Wald 4B v2.1
  136. 68.9 JevOne
  137. 68.8 vLLM-SR d3-mini
  138. 68.7 Torchcast Decision 4B (v1.1)
  139. 68.6 iota-decide-4b-v0.3 (Macrocosmos, letter-readout 4B, self-hosted)
  140. 68.6 Xevn-4B
  141. 68.5 Chon-1.0-E4B
  142. 68.3 Strike-1
  143. 68.3 DecisionNode Flash
  144. 68.3 SimpleJev Qwen3.6-35B-A3B
  145. 68.2 Hopper 12B trained (gemma-4-12B-it frozen + LoRA r32 unmerged, one forward pass, self-hosted)
  146. 68.1 SPX-CD-Omni (SurdAI, google/gemma-4-12B-it + LoRA r32, one forward pass, self-hosted)
  147. 68.0 NInfer Qwen3.8-27B NVFP4
  148. 68.0 Open-Jev 27B v1.1 (Zefan Cai, LoRA + decision head on Qwen3.8-27B)
  149. 67.1 Aplomb 1 (5.3B decision model on Qwen3.5-4B, trained readout head, self-hosted)
  150. 66.8 Bespoke Nimble 9B v3 (Qwen3.5-9B LoRA)
  151. 66.7 Amazon Strands Decider 26B-A4B
  152. 66.6 LitJev (Qwen3.8-27B)
  153. 66.4 Saif 4B (Gyr Technology)
  154. 66.4 Tern-35B-A3B (matthewagi, Qwen3.6 35B-A3B MoE decision model, author server, self-hosted)
  155. 66.2 KnowLine 4B-Gen5.5
  156. 66.2 decider-35b-a3b (Mapika)
  157. 66.1 Bev / Bonsai 27B
  158. 66.0 Metask rain 4B
  159. 65.9 cribl-decision-1.0-9B (Cribl)
  160. 65.5 shisa-de-1 (shisa-ai, Gemma-4-26B-A4B)
  161. 65.3 Plumb-4B (crh225, JevK5 v0.2 + LoRA)
  162. 65.1 cribl-decision-2 (Cribl)
  163. 65.0 Arcbright Elf 1 12B
  164. 64.9 decider-4b v2 (Mapika)
  165. 64.8 Seb-9B-v2 (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted)
  166. 64.5 Granite Switch 4.1 30B (revision 4.1.1, decision adapter)
  167. 64.3 HS1 0.1.2 (Hyland)
  168. 64.2 JevK5 v0.3 (4B)
  169. 64.0 Nagi-HUGE 12B (nagisanzenin, Gemma-4-12B LoRA)
  170. 64.0 Bespoke Nimble-v4 (Bespoke Labs)
  171. 64.0 Jeffan-4B-2010
  172. 63.8 Clef-Flash (Cloudflare, Qwen3.5-9B post-train with a joint schema head, multimodal, measured on text)
  173. 63.8 JPT-4B (kirp, Qwen3.5-4B LoRA)
  174. 63.7 Vansa-3.4 (Vansa, hosted System One API)
  175. 63.6 deck-4B v1.0 (krishna765, JevK5 + LoRA, FP8 weight-only)
  176. 63.4 Jebadiah 9B v2.1 (Frontier Infra, Qwen3.5-9B LoRA)
  177. 63.2 GLiClass v4 9B (Knowledgator)
  178. 63.1 Seb-9B (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted)
  179. 63.0 wity-1 (Wity, reasoning off)
  180. 62.9 Amos 1 medium (Scira)
  181. 62.9 janus 4B (Icarus AI / cmxu, Qwen3.5-4B + LoRA r64 and pointer decision head)
  182. 62.8 JevK5 v0.2.0
  183. 62.4 Messier One v0.4 (Qwen3.5-4B fine-tune, self-hosted)
  184. 62.4 Imajev-9B (mohit67890, Qwen3.5-9B LoRA)
  185. 61.9 Imajev-4B (RTX 5090)
  186. 61.9 jqv (Qwen3-32B zero-shot)
  187. 61.9 Open-Jev 9B (Zefan Cai)
  188. 61.8 metask-jev-4b
  189. 61.7 Kahn1 4B (Okura66, Qwen3.5-4B LoRA merge, author's sysone engine)
  190. 61.7 pplx-decider-flash-v1.2 (Perplexity, Qwen3.5-4B full fine-tune, marker readout, self-hosted)
  191. 61.7 lev (Interfaze AI, Qwen3.5-4B + LoRA r32, label-token readout + candidate-path head, per-bucket temperatures)
  192. 61.6 Decision 2.0 Lux 9B (vLLM-SR, Qwen3.5-9B lineage)
  193. 61.6 JEB-35B-A3B (szybkie.ai, Qwen3.6-35B-A3B LoRA)
  194. 61.5 Kev-9B v2 (Jared Palmer, Qwen3.5-9B-Base LoRA + pointer head)
  195. 61.5 Decisio v0.9.0 on Qwen3.6-35B-A3B (official FP8, self-hosted)
  196. 61.4 blink-4b (thegovind, Qwen3.5-4B derivative)
  197. 61.2 Jev-Style-2B
  198. 61.2 Hopper
  199. 61.1 Malkuth-4B (newfull5, Kev post-train)
  200. 60.9 Qwen3.5-9B Jev-like data-mix v2
  201. 60.4 BB-Qwen3.5-4B-LoRA (Babak Barazandeh, LoRA on Qwen3.5-4B-Base, hosted /v1/systemone)
  202. 60.4 Decision 4B v1.2 (FlyMyJev, Qwen3.5-4B + LoRA)
  203. 60.4 APUS-OpenJev-v1-35B-A3B (merged bf16 checkpoint-5949 MoE on Qwen3.5-35B-A3B, their own native runtime, full depth, self-hosted)
  204. 60.2 jiwo 4b (Qwen3.5-4B backbone with a trained decision head, one forward pass, native probabilities, self-hosted)
  205. 60.1 PurpleMIST-Flash-1.0 (Olaverse)
  206. 60.0 Standard One 8B SH (Standard Thinking)
  207. 59.8 Gemini 3.1 Flash-Lite
  208. 59.6 Verdin-4B (Victor Bennett, Qwen3.5-4B + merged LoRA r16 + pointer head)
  209. 59.6 NVIDIA Nemotron 3.5 Super VL
  210. 59.5 Wald 4B
  211. 59.3 Quyet-1.0-Medium (Chinh Nguyen, Qwen3.5-4B decoder, option-letter logits)
  212. 59.1 Decision-4B (Eval Engine / Chromia)
  213. 58.9 reflex 4B (kshetrajna12)
  214. 58.7 openjev-sglang (Qwen3.6-35B-A3B on SGLang)
  215. 58.4 Standard One 8B (Standard Thinking)
  216. 58.3 Clef-omni
  217. 58.3 Manchego v2.1
  218. 58.1 Messier One v0.2 (Qwen3.5-4B fine-tune, one prefill, self-hosted)
  219. 58.1 Diffusion Jev (DiffusionGemma 26B-A4B on patched SGLang, 48-step diffusion readout, self-hosted)
  220. 57.8 baselabs joey-1 (2B)
  221. 57.5 jev-local (Qwen3.5-9B)
  222. 57.2 Instinct Dual 4B
  223. 57.2 Decision 4B v1.1 (FlyMyJev, Qwen3.5-4B + LoRA)
  224. 57.1 Granite Switch 4.1 8B (revision 4.1.1, decision adapter)
  225. 57.1 JEV Qwen3.5-9B Base NVFP4
  226. 57.1 iota-decide-4b (Macrocosmos, Clef joint-head 4B, self-hosted)
  227. 57.1 Uniphore-4B-Decider
  228. 57.0 Trio-Spark v1.3 (MachineFi)
  229. 57.0 Jebadiah 9B v2 (Frontier Infra, Qwen3.5-9B LoRA)
  230. 56.9 CoCo-Decision-4B (corners-ai, LoRA on Qwen3.5-4B, option-label logits of one pass, served by oh-my-jev)
  231. 56.9 APUS-OpenJev-v1-9B (merged bf16 checkpoint-3000 on Qwen3.5-9B, their own native runtime, full depth, self-hosted)
  232. 56.8 Decider-4B (roudone, NeoHorse 4B pointer model, author server, self-hosted)
  233. 56.5 typecastlm (Mikhail Gribov, Qwen3.5-4B computed head)
  234. 56.5 Bespoke Nimble 9B (Bespoke Labs)
  235. 56.5 pjev-2b (Kailune AI, frozen MiniCPM5-2B + merged LoRA r64, letter-logit readout, T=1 stub)
  236. 56.4 TypeCastLM 1.4.0 (Mikhail Gribov, Qwen3.5-4B computed head)
  237. 56.3 Sors-2B (SakuraYuyuko)
  238. 56.3 RYOTIDE-Qwen (csabag, Qwen3.5-4B stock weights)
  239. 56.1 Amber 4B (MLukauskas)
  240. 56.1 Compass 0.2.0 (adimyth, Qwen3.5-4B LoRA)
  241. 56.0 Standard One 3B SH v4 (Standard Thinking)
  242. 55.6 Standard One 3B SH (Standard Thinking)
  243. 55.4 SemIf, formerly OpenJev (Qwen3.5-4B, TheoLeeCJ)
  244. 55.1 ArseneLupin v1.1 (IBOU-SEARCH, Qwen3.5-4B LoRA)
  245. 55.0 local-jev Qwen3.5-4B
  246. 54.8 Decision 2B (FlyMy.AI, v59)
  247. 54.8 Bongard-mini (AgentBull, T5Gemma-2 4B-4B)
  248. 54.5 RYOTIDE-Gemma (csabag, Gemma-4-E4B stock weights)
  249. 54.4 Jez-large-1.0.0 (pragmaticfrontier, Gemma-4-E4B label-logit decision model, author server, self-hosted)
  250. 54.3 Noma (Blackdrome AI Labs, Qwen3.5-4B-Base derivative)
  251. 54.0 spark-s1-4b-v6 (Open Spark Jev, abhishek085)
  252. 54.0 Imajev-2B (mohit67890, Qwen3.5-2B LoRA)
  253. 53.2 cribl-decision-2-mini (Cribl)
  254. 52.5 RYOTIDE-Qwen9
  255. 52.3 classify-1 (ScaleDown)
  256. 52.3 Arbiter-4B (Goddragon, NeoHorse 4B pointer model, author server, self-hosted)
  257. 52.1 DecisionNode Edge 0.9.2
  258. 51.9 system-one-open (Gemma 4 E2B LoRA on an L4)
  259. 51.1 D1A on Gemma 4 E4B (JohnP1 d1a-e4b)
  260. 50.5 ClassOne Qwen 3.5 9B (Qwen3.5-9B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted)
  261. 50.0 kev 8B (research preview)
  262. 49.9 Nev-2B (EryriLabs)
  263. 49.5 Liquid AI d1-3B
  264. 49.4 ZeroEntropy zerank-2
  265. 49.3 JADE (Qwen3.8-27B LoRA)
  266. 49.1 open-alternative-jev (Qwen3.5-4B, IkerMoel)
  267. 48.6 Qwen3.8 27B (Chutes TEE)
  268. 48.6 OpenSourceJev (Qwen3.5-4B Q4_K_M, native llama.cpp)
  269. 47.7 Open-Jev 2B (Zefan Cai)
  270. 47.7 EXAONE-4.0-1.2B-JEV v0.3 (unofficial fine-tune of LGAI-EXAONE/EXAONE-4.0-1.2B, native option-label softmax, two reads averaged, self-hosted)
  271. 47.6 Malkuth-2B (newfull5, Kev post-train)
  272. 47.6 DecisionNode Edge Mini 0.9.2
  273. 47.5 Nemotron Diffusion 8B (pst2154, optimized vLLM)
  274. 47.2 jiwo 0.8b (Qwen3.5-0.8B backbone with a trained decision head, one forward pass, native probabilities, self-hosted)
  275. 46.1 decision-machine-1 (milliseconds.ai)
  276. 45.7 decisor-4b
  277. 45.5 D1A on Gemma 4 E2B (JohnP1 d1a-e2b)
  278. 44.8 Qwen3-Reranker-4B
  279. 44.6 decider-2b (Mapika)
  280. 44.2 Laya typed-decisions
  281. 44.1 BAAI bge-reranker-v2-m3
  282. 44.0 kev 4B (research preview)
  283. 43.9 Mixedbread mxbai-rerank-base-v2
  284. 43.4 lev-350m (Franck Verrot, LFM2.5-350M)
  285. 43.2 Jez-small-1.0.0 (pragmaticfrontier, Qwen3.5-0.8B label-logit decision model, author server, self-hosted)
  286. 43.2 Certo v1 (AltSlate Labs)
  287. 43.0 PurpleMIST-Mini-1.0 (Olaverse)
  288. 43.0 Chon-1.0-12B
  289. 43.0 smalljev semantic-v9
  290. 42.6 Vega-U (Qwen3.5-4B reader)
  291. 42.6 jul fast (usejul/jul-decision-minicpm5-2b: MiniCPM5-2B fine-tune with a pointer head, option probabilities from one pass, self-hosted via `jul serve`)
  292. 42.6 Qwen3.5-0.8B Decision Model (Mourad Ghafiri)
  293. 42.4 Krino-2B
  294. 42.3 Gutsy 0.8B v0.3
  295. 41.7 Alibaba GTE Reranker ModernBERT-base
  296. 41.7 Jeetu Decider 2B
  297. 41.6 watt-flash-0.1 (Zaitgeist Labs, 140M mmBERT-small encoder with an option-marker scorer, one forward pass per request, self-hosted)
  298. 40.6 Decision Fast (FlyMy.AI, v53a)
  299. 40.4 openJev Verdict 1.4
  300. 39.5 Quyet-1.0-Small-EN (Chinh Nguyen, ModernBERT-base encoder with fixed heads)
  301. 38.6 Reflex Instinct 0.6B (Atlas AI)
  302. 38.4 ClassOne Gemma 4 E4B (Gemma-4-E4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted)
  303. 38.1 ClassOne Qwen 3.5 4B (Qwen3.5-4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted)
  304. 38.0 OpenDecision (ModernBERT-large zero-shot)
  305. 37.6 verdict-small (Manavarya09, multilingual-e5-small 118M)
  306. 36.9 kev 0.6B (research preview)
  307. 36.9 jevora-9b-v1 (kadirbekovvv / doxanocap, Qwen3.5-9B-Base derivative)
  308. 36.6 Vega 4B
  309. 36.1 Vega-U (Qwen3.5-0.8B reader)
  310. 35.5 WaterSheep (Samrat Dutta, ModernBERT-base encoder with calibrated heads, served by the author's own `watersheep --serve`)
  311. 35.4 open-jev-deberta-v3-large (local CPU)
  312. 35.3 Raw Phi-4 mini direct logits
  313. 34.6 modernbert-jev v2
  314. 34.6 Fastino GLiNER-2.5-Decide (hosted API)
  315. 34.6 Raw Qwen3 4B Instruct 2507 direct logits
  316. 33.0 Vega 0.8B
  317. 32.9 kev 0.5B
  318. 32.8 Liquid AI d1-omni-600M
  319. 32.5 Laya (Convai Innovations, ModernBERT-large 421M)
  320. 32.5 Quyet-1.0-Small (Chinh Nguyen, SEA-LION-ModernBERT-300M encoder with fixed heads)
  321. 31.1 Quyet-1.0-Tiny (Chinh Nguyen, mmBERT-small encoder (16 layers kept) with fixed heads)
  322. 31.1 SimpleJev (Qwen3.5-0.8B, CPU)
  323. 30.9 TDE general large v0.1 (tdelab, ModernBERT-large encoder)
  324. 30.6 Open Jev JSON Canvas (JoshuaSP)
  325. 30.2 Phocinae-Largha-150M-v1
  326. 30.1 Raw Qwen3 8B direct logits
  327. 29.7 system-one (Qwen3-8B, Sean Goedecke)
  328. 28.8 JevAct (einptein, jev1-2b-v2)
  329. 26.6 openJev Verdict (heman10x, ModernBERT-base 151M)
  330. 25.8 Mirror
  331. 24.2 Tacet Sonata (CodePawl, 144M packed encoder on mmBERT-small, option-marker softmax in one pass, in-process via the author's tacet package)
  332. 20.1 CLM-8B (Contrastive-LM, clm-latest)
  333. 19.2 ClassOne Qwen 3.5 2B (Qwen3.5-2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted)
  334. 18.1 Deem 0.8B v1
  335. 17.0 Laya multilingual
  336. 15.2 ClassOne Gemma 4 E2B (Gemma-4-E2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted)
  337. 9.85 Julia-1 (Supersonic Labs)
  338. 8.32 Raw Qwen3 1.7B direct logits
  339. 8.25 Raw Qwen3 0.6B direct logits
  1. 1 GPT-6 Luna v1.8.9; api; outside capability caps; ranked; tariff 98.4%
  2. 2 GPT-5.6 Luna v1.8.9; api; outside capability caps; ranked; tariff 93.6%
  3. 3 Torchcast Decision 31B v1.0 v1.8.9; open; outside capability caps; unranked; estimate 91.8%
  4. 4 MetaDecision-31B-Rain v1.8.9; open; outside capability caps; unranked; estimate 90.7%
  5. 5 DeepSeek V4.1 Flash v1.8.9; api; outside capability caps; ranked; tariff 89.6%
  6. 6 Karpo-Decision-31B v1.8.9; open; outside capability caps; unranked; estimate 89.5%
  7. 7 WeMM-JEV (Rover00) v1.8.9; open; outside capability caps; unranked; estimate 88.7%
  8. 8 Jeff-1.1-Large (jgeuter) v1.8.9; open; outside capability caps; unranked; estimate 88.2%
  9. 9 VisionJEV 1.0.0 v1.8.9; open; outside capability caps; unranked; estimate 86.5%
  10. 10 kettei-fat (Sqwish) v1.8.9; api; capability ranked; ranked; estimate 86.3%
  11. 11 Piccolo 27B (agentscope-ai) v1.8.9; open; outside capability caps; unranked; estimate 86%
  12. 12 Super Decider (thinking) v1.8.9; api; capability ranked; ranked; listed 85.8%
  13. 13 baselabs mammoth-1 (27B) v1.8.9; open; capability ranked; ranked; estimate 85.2%
  14. 14 KnowLine 31B-Gen1 v1.8.9; open; capability ranked; ranked; estimate 85.2%
  15. 15 Fastino GLiDE 1.2 (open weights, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 84.9%
  16. 16 Super Decider v1.8.9; api; capability ranked; ranked; listed 84.9%
  17. 17 Fastino GLiDE (hosted API, 22 Oct 2026) v1.8.9; api; capability ranked; ranked; mixed 84.5%
  18. 18 H2O-Lightning-31B v1.0.7 (H2O.ai, Gemma-4-31B-it fine-tune, vLLM + h2oai overlay, one-engine FP8 profile, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 84.3%
  19. 19 Drex Max (Nace AI) v1.8.9; api; capability ranked; ranked; listed 84.2%
  20. 20 decide-1.1 (decide.sh, API) v1.8.9; api; outside capability caps; unranked; tariff 84%
  21. 21 oenogarum (plantwaterer; commercially restricted) v1.8.9; open; outside capability caps; unranked; estimate 83.9%
  22. 22 allec-oavg (plantwaterer; commercially restricted) v1.8.9; open; outside capability caps; unranked; estimate 83.2%
  23. 23 Perplexity Decider v1.1 27B (Qwen3.8-27B, noncausal decision readout) v1.8.9; open; capability ranked; ranked; estimate 82.7%
  24. 24 vLLM-SR d3-turbo v1.8.9; open; capability ranked; ranked; estimate 82.6%
  25. 25 StartLux-Decision v1.1 27B Instant (StartLux Labs, Qwen3.8-27B fine-tune) v1.8.9; open; capability ranked; ranked; estimate 82.6%
  26. 26 Decision-King v1.1 (xsdfasfgsa) v1.8.9; open; outside capability caps; unranked; estimate 82.3%
  27. 27 vLLM-SR d3 v1.8.9; open; capability ranked; ranked; estimate 82.2%
  28. 28 basal-omni-1.0-ultra (Remek Kinas, Gemma-4-31B-it + merged LoRA, typed-decision readout, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 82.1%
  29. 29 DOPEST 1.0 31B v1.8.9; open; capability ranked; ranked; estimate 82%
  30. 30 SubQ Snap 1.1 (SubQ) v1.8.9; api; capability ranked; ranked; estimate 81.9%
  31. 31 Jeff-1.1-Medium (jgeuter) v1.8.9; open; capability ranked; ranked; estimate 81.9%
  32. 32 Arcbright Elf 1 31B v1.8.9; open; capability ranked; ranked; estimate 81.8%
  33. 33 Quyet-1.0-Large (Chinh Nguyen, Gemma-4-31B decoder, option-letter logits) v1.8.9; open; capability ranked; ranked; estimate 81.7%
  34. 34 Jeffan-31b-1710 v1.8.9; open; capability ranked; ranked; estimate 81.7%
  35. 35 Jeff-1.0-Large v1.8.9; open; capability ranked; ranked; estimate 81.5%
  36. 36 SubQ Snap (preview, SubQ) v1.8.9; api; capability ranked; ranked; estimate 81.5%
  37. 37 AJev lm6q (Gemma 4 31B) v1.8.9; open; outside capability caps; ranked; estimate 81.4%
  38. 38 Nimble-v4-max (open weights, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 81.2%
  39. 39 Sieve-31B-v2 v1.8.9; open; capability ranked; ranked; estimate 81.1%
  40. 40 Turion Jev 31B v1.8.9; open; capability ranked; ranked; estimate 80.9%
  41. 41 pplx-decider-v1.2 (Perplexity, Qwen3.8-27B full fine-tune, marker readout, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 80.8%
  42. 42 d1-ultra (Liquid AI) v1.8.9; api; capability ranked; ranked; estimate 80.5%
  43. 43 torchcast-decision-27b (Torchcast AI, Qwen3.8-27B fine-tune) v1.8.9; open; capability ranked; ranked; estimate 80.2%
  44. 44 Arcbright Elf 1 12B (r2) v1.8.9; open; outside capability caps; unranked; estimate 79.9%
  45. 45 Lychee-27B v1.8.9; open; capability ranked; ranked; estimate 79.8%
  46. 46 Microsoft-Decision-1 (updated weights, Oct 2026) v1.8.9; api; outside capability caps; unranked; tariff 79.8%
  47. 47 Bespoke Nimble-v4-max (Bespoke Labs) v1.8.9; api; capability ranked; ranked; estimate 79.7%
  48. 48 Jeff-1.1-Qwen-4B (jgeuter) v1.8.9; open; outside capability caps; unranked; estimate 79.6%
  49. 49 metask-jev-rain-31B-trained service v3 v1.8.9; open; capability ranked; ranked; estimate 79.6%
  50. 50 decisio v0.8.0 on gemma-4-31B-it (frozen, FP8 on load, one prefill per state, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 79.6%
  51. 51 Vera Large v1.8.9; open; capability ranked; ranked; estimate 79.5%
  52. 52 Kev 27B (Kev 1.0, Qwen3.8-27B) v1.8.9; open; capability ranked; ranked; estimate 78.9%
  53. 53 metask-jev-rain-31B-trained v1.8.9; open; capability ranked; ranked; estimate 78.9%
  54. 54 Myles-G 31B (Launchables) v1.8.9; api; capability ranked; ranked; estimate 78.8%
  55. 55 StartLux-Decision v1.1 27B AutoThink (StartLux Labs, Qwen3.8-27B fine-tune, gate 0.6) v1.8.9; open; capability ranked; ranked; estimate 78.8%
  56. 56 Decisio 12B tuned commit-think (variant of the Decisio family) v1.8.9; open; outside capability caps; unranked; estimate 78.7%
  57. 57 Sage 1.3.0 (Levanto Labs) v1.8.9; api; capability ranked; ranked; estimate 78.6%
  58. 58 SPX-CD Pro (SurdAI, hosted /v1/systemone, Oct-4 checkpoint) v1.8.9; api; outside capability caps; ranked; estimate 78.5%
  59. 59 metask-jev-rain-12B v1.8.9; open; outside capability caps; wrapper; estimate 78.3%
  60. 60 classifier.dev (fast tier) v1.8.9; api; outside capability caps; wrapper; estimate 78.3%
  61. 61 Amos 1 large (Scira) v1.8.9; api; outside capability caps; ranked; estimate 78%
  62. 62 Bobcat 1.3 (Gemma 4 31B) v1.8.9; open; capability ranked; ranked; estimate 77.9%
  63. 63 deck-31B (krishna765, frozen Gemma-4-31B-it, TorchAO FP8 dynamic) v1.8.9; open; capability ranked; ranked; estimate 77.6%
  64. 64 SD-JEV-4B-v12.1.0 (Kego, author-hosted API) v1.8.9; api; outside capability caps; unranked; estimate 77.5%
  65. 65 Torchcast Decision 4B (v1.3) v1.8.9; open; capability ranked; ranked; estimate 77.5%
  66. 66 KnowLine 26B-A4B-Gen2.5 v1.8.9; open; capability ranked; ranked; estimate 77.3%
  67. 67 Eikos-27B (caiovicentino1, Qwen3.8-27B) v1.8.9; open; capability ranked; ranked; estimate 77.2%
  68. 68 Jev 1.13.0 (TypeSafe AI) v1.8.9; reference; capability ranked; ranked; tariff 77.1%
  69. 69 wity-1 (Wity, reasoning always) v1.8.9; api; outside capability caps; listed; tariff 77.1%
  70. 70 H2O-Lightning-4B v1.4.1 v1.8.9; open; outside capability caps; unranked; estimate 77%
  71. 71 Fastino GLiDE v1.8.9; api; outside capability caps; ranked; tariff 77%
  72. 72 wity-1 (Wity, reasoning auto) v1.8.9; api; outside capability caps; ranked; tariff 76.9%
  73. 73 metask-jev-rain-12B-trained v1.8.9; open; capability ranked; ranked; estimate 76.9%
  74. 74 Aleria Judge 1.1.1 (ALERIA-ML, inference technique on stock Gemma 4 31B, author server, self-hosted) v1.8.9; open; outside capability caps; ranked; estimate 76.9%
  75. 75 Wald 26B-A4B (Gemma 4 26B-A4B) v1.8.9; open; capability ranked; ranked; estimate 76.8%
  76. 76 DecisionNode v1.8.9; api; outside capability caps; unranked; unpriced 76.8%
  77. 77 KnowLine 35B-A3B-Gen1.5 v1.8.9; open; capability ranked; ranked; estimate 76.7%
  78. 78 Piccolo 9B (agentscope-ai) v1.8.9; open; outside capability caps; unranked; estimate 76.6%
  79. 79 cribl-decision-1.0-27B (Cribl) v1.8.9; open; outside capability caps; unranked; estimate 76.5%
  80. 80 René-1 31B FP8 (salfatigroup, Gemma 4 31B full fine-tune, one-pass option readout) v1.8.9; open; capability ranked; ranked; estimate 76%
  81. 81 OpenJev (thinking, BF16) v1.8.9; open; outside capability caps; ranked; estimate 75.8%
  82. 82 Decisio 12B tuned commit (variant of the Decisio family) v1.8.9; open; capability ranked; ranked; estimate 75.7%
  83. 83 Clef (Cloudflare, Qwen3.8-27B post-train with a joint schema head, multimodal, measured on text) v1.8.9; open; capability ranked; ranked; estimate 75.3%
  84. 84 Jev 1.14.0 (TypeSafe AI) v1.8.9; api; capability ranked; ranked; estimate 75%
  85. 85 H2O-Lightning-4B v1.1 (H2O.ai, Qwen3.5-4B fine-tune, stock vLLM + open shim) v1.8.9; open; capability ranked; ranked; estimate 75%
  86. 86 SimpleJev Qwen3.8-27B (self-hosted, v1.6 pool) v1.8.9; open; outside capability caps; ranked; estimate 74.9%
  87. 87 Umpire 35B-A3B v1.8.9; open; outside capability caps; unranked; estimate 74.5%
  88. 88 z3-mini-soup-preview (Zils AI) v1.8.9; open; outside capability caps; unranked; estimate 74.5%
  89. 89 d1 (Liquid AI) v1.8.9; api; capability ranked; ranked; tariff 74.4%
  90. 90 SPX-CD Flash (SurdAI, hosted /v1/systemone, Oct-4 checkpoint) v1.8.9; api; capability ranked; ranked; estimate 74.3%
  91. 91 iota-decide-4b-v0.4 (Macrocosmos, letter-readout 4B, self-hosted) v1.8.9; open; outside capability caps; unranked; estimate 74%
  92. 92 JEV-27B (AutoTrust, Qwen3.8-27B) v1.8.9; open; capability ranked; ranked; estimate 73.9%
  93. 93 Nagi-ENORMOUS 27B (nagisanzenin, Qwen3.8-27B LoRA) v1.8.9; open; capability ranked; ranked; estimate 73.8%
  94. 94 Mercury Decide (Inception; System One decisions API, served free on OpenRouter as inception/mercury-decide:free) v1.8.9; api; capability ranked; ranked; estimate 73.6%
  95. 95 Celeris-1 Decision (Celeris) v1.8.9; api; capability ranked; ranked; estimate 73.6%
  96. 96 Piccolo 4B (agentscope-ai) v1.8.9; open; outside capability caps; unranked; estimate 73.6%
  97. 97 OpenAI Decisions (gpt-6-luna) v1.8.9; api; capability ranked; ranked; tariff 73.5%
  98. 98 Surogate Rune 26B-A4B v3 (v1.6 pool) v1.8.9; open; capability ranked; ranked; estimate 73.5%
  99. 99 RSI-Jev v6.1-VL 27B v1.8.9; open; capability ranked; ranked; estimate 73.5%
  100. 100 Sage 1.4 (Levanto Labs) v1.8.9; api; outside capability caps; ranked; estimate 73.3%
  101. 101 kettei-chub (Sqwish) v1.8.9; api; capability ranked; ranked; estimate 73.1%
  102. 102 Bobcat Flash 1.2 (Gemma-4-26B-A4B-it + two merged rank-64 LoRA adapters, typed-decision readout at the first answer position, FP8, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 73%
  103. 103 Autoloops – Gemma 4 31B IT v1.8.9; api; outside capability caps; ranked; tariff 72.7%
  104. 104 Torchcast Decision 4B (v1.2) v1.8.9; open; capability ranked; ranked; estimate 72.6%
  105. 105 swanOne (blockbrain, Qwen3.8-Flash-Next NVFP4) v1.8.9; open; capability ranked; ranked; estimate 72.4%
  106. 106 FrontiersMind Lumma-fev-27B v1.8.9; open; capability ranked; ranked; estimate 72.3%
  107. 107 Microsoft-Decision-1 (Azure Foundry) — A2 re-run v1.8.9; api; outside capability caps; preliminary; tariff 72%
  108. 108 torchcast-decision-12b (Torchcast AI, Gemma-4-12B fine-tune, option-letter logprob readout) v1.8.9; open; capability ranked; ranked; estimate 71.7%
  109. 109 AutoJev-27B (denis-pplx, Qwen3.8-27B) v1.8.9; open; capability ranked; ranked; estimate 71.6%
  110. 110 SimpleJev Qwen3.8-27B v1.8.9; api; outside capability caps; ranked; estimate 71.6%
  111. 111 decider-12b v2 (Mapika) v1.8.9; open; capability ranked; ranked; estimate 71.3%
  112. 112 Decision 2.0 Vega 27B (vLLM-SR, Qwen3.8-27B) v1.8.9; open; capability ranked; ranked; estimate 71.3%
  113. 113 Jev-Omni (akhilaaa3, Gemma-4-12B merged) v1.8.9; open; capability ranked; ranked; estimate 71.3%
  114. 114 Winnow-12B Q8 v1.8.9; open; capability ranked; ranked; estimate 71.3%
  115. 115 Xor 26B-A4B (Juspay, Gemma-4-26B-A4B, bf16) v1.8.9; open; capability ranked; ranked; estimate 71.2%
  116. 116 decider-12b v1, stock Gemma-4-12B-it (Mapika) v1.8.9; open; capability ranked; ranked; estimate 71%
  117. 117 Cygnet (blockbrain, frozen Gemma-4-12B-it) v1.8.9; open; capability ranked; ranked; estimate 70.9%
  118. 118 Microsoft-Decision-1 (Azure Foundry) v1.8.9; api; capability ranked; ranked; tariff 70.8%
  119. 119 decisio v0.8.0 on gemma-4-12B-it (frozen, one forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 70.7%
  120. 120 JPT-35B-A3B (Qwen3.5-35B-A3B fine-tune) v1.8.9; open; capability ranked; ranked; estimate 70.7%
  121. 121 Myles-G 12B (Launchables) v1.8.9; api; capability ranked; ranked; estimate 70.7%
  122. 122 Instinct (ZooWork, Qwen3.8-27B) v1.8.9; api; capability ranked; ranked; tariff 70.5%
  123. 123 decider chat on Gemma-4-31B-it (Mapika, frozen base, inference technique) v1.8.9; open; capability ranked; ranked; estimate 70.5%
  124. 124 baselabs elephant-1 (9B) v1.8.9; open; capability ranked; ranked; estimate 70.3%
  125. 125 Jebadiah 27B (Frontier Infra, Qwen3.8-27B LoRA) v1.8.9; open; capability ranked; ranked; estimate 70.2%
  126. 126 GEV-26B-Decide (AutoTrust, Gemma-4-26B-A4B + LoRA + head) v1.8.9; open; capability ranked; ranked; estimate 70.1%
  127. 127 Chon-1.0-12B-Fast v1.8.9; open; capability ranked; ranked; estimate 70.1%
  128. 128 Blink v0.3 26B-A4B v1.8.9; open; capability ranked; ranked; estimate 70%
  129. 129 decisio v0.9.0 on gemma-4-12B-it (frozen, one prefill per question, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 70%
  130. 130 NInfer Qwen3.8-Flash-Next mixed v1.8.9; open; capability ranked; ranked; estimate 69.7%
  131. 131 attest-31B-v1.0 v1.8.9; api; capability ranked; ranked; estimate 69.5%
  132. 132 Respan Span-02 v1.8.9; open; capability ranked; ranked; estimate 69.3%
  133. 133 Nemotron Deciffusion 8B v2 (pst2154) v1.8.9; open; outside capability caps; listed; unpriced 69.3%
  134. 134 Philly Decider v2 (PhillyLabs) v1.8.9; open; capability ranked; ranked; estimate 69.3%
  135. 135 Wald 4B v2.1 v1.8.9; open; capability ranked; ranked; estimate 69.3%
  136. 136 JevOne v1.8.9; open; capability ranked; ranked; estimate 68.9%
  137. 137 vLLM-SR d3-mini v1.8.9; open; capability ranked; ranked; estimate 68.8%
  138. 138 Torchcast Decision 4B (v1.1) v1.8.9; open; capability ranked; ranked; estimate 68.7%
  139. 139 iota-decide-4b-v0.3 (Macrocosmos, letter-readout 4B, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 68.6%
  140. 140 Xevn-4B v1.8.9; open; capability ranked; ranked; estimate 68.6%
  141. 141 Chon-1.0-E4B v1.8.9; open; capability ranked; ranked; estimate 68.5%
  142. 142 Strike-1 v1.8.9; api; outside capability caps; unranked; estimate 68.3%
  143. 143 DecisionNode Flash v1.8.9; api; outside capability caps; unranked; unpriced 68.3%
  144. 144 SimpleJev Qwen3.6-35B-A3B v1.8.9; api; outside capability caps; ranked; estimate 68.3%
  145. 145 Hopper 12B trained (gemma-4-12B-it frozen + LoRA r32 unmerged, one forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 68.2%
  146. 146 SPX-CD-Omni (SurdAI, google/gemma-4-12B-it + LoRA r32, one forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 68.1%
  147. 147 NInfer Qwen3.8-27B NVFP4 v1.8.9; open; capability ranked; ranked; estimate 68%
  148. 148 Open-Jev 27B v1.1 (Zefan Cai, LoRA + decision head on Qwen3.8-27B) v1.8.9; open; outside capability caps; ranked; estimate 68%
  149. 149 Aplomb 1 (5.3B decision model on Qwen3.5-4B, trained readout head, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 67.1%
  150. 150 Bespoke Nimble 9B v3 (Qwen3.5-9B LoRA) v1.8.9; open; capability ranked; ranked; estimate 66.8%
  151. 151 Amazon Strands Decider 26B-A4B v1.8.9; open; capability ranked; ranked; estimate 66.7%
  152. 152 LitJev (Qwen3.8-27B) v1.8.9; open; outside capability caps; ranked; estimate 66.6%
  153. 153 Saif 4B (Gyr Technology) v1.8.9; open; outside capability caps; unranked; estimate 66.4%
  154. 154 Tern-35B-A3B (matthewagi, Qwen3.6 35B-A3B MoE decision model, author server, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 66.4%
  155. 155 KnowLine 4B-Gen5.5 v1.8.9; open; capability ranked; ranked; estimate 66.2%
  156. 156 decider-35b-a3b (Mapika) v1.8.9; open; capability ranked; ranked; estimate 66.2%
  157. 157 Bev / Bonsai 27B v1.8.9; open; outside capability caps; ranked; estimate 66.1%
  158. 158 Metask rain 4B v1.8.9; open; capability ranked; ranked; estimate 66%
  159. 159 cribl-decision-1.0-9B (Cribl) v1.8.9; open; outside capability caps; unranked; estimate 65.9%
  160. 160 shisa-de-1 (shisa-ai, Gemma-4-26B-A4B) v1.8.9; open; capability ranked; ranked; estimate 65.5%
  161. 161 Plumb-4B (crh225, JevK5 v0.2 + LoRA) v1.8.9; open; capability ranked; ranked; estimate 65.3%
  162. 162 cribl-decision-2 (Cribl) v1.8.9; open; outside capability caps; unranked; estimate 65.1%
  163. 163 Arcbright Elf 1 12B v1.8.9; open; capability ranked; ranked; estimate 65%
  164. 164 decider-4b v2 (Mapika) v1.8.9; open; capability ranked; ranked; estimate 64.9%
  165. 165 Seb-9B-v2 (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 64.8%
  166. 166 Granite Switch 4.1 30B (revision 4.1.1, decision adapter) v1.8.9; open; outside capability caps; unranked; estimate 64.5%
  167. 167 HS1 0.1.2 (Hyland) v1.8.9; api; capability ranked; ranked; estimate 64.3%
  168. 168 JevK5 v0.3 (4B) v1.8.9; open; capability ranked; ranked; estimate 64.2%
  169. 169 Nagi-HUGE 12B (nagisanzenin, Gemma-4-12B LoRA) v1.8.9; open; outside capability caps; listed; unpriced 64%
  170. 170 Bespoke Nimble-v4 (Bespoke Labs) v1.8.9; api; capability ranked; ranked; estimate 64%
  171. 171 Jeffan-4B-2010 v1.8.9; open; capability ranked; ranked; estimate 64%
  172. 172 Clef-Flash (Cloudflare, Qwen3.5-9B post-train with a joint schema head, multimodal, measured on text) v1.8.9; open; capability ranked; ranked; estimate 63.8%
  173. 173 JPT-4B (kirp, Qwen3.5-4B LoRA) v1.8.9; open; capability ranked; ranked; estimate 63.8%
  174. 174 Vansa-3.4 (Vansa, hosted System One API) v1.8.9; api; capability ranked; ranked; tariff 63.7%
  175. 175 deck-4B v1.0 (krishna765, JevK5 + LoRA, FP8 weight-only) v1.8.9; open; capability ranked; ranked; estimate 63.6%
  176. 176 Jebadiah 9B v2.1 (Frontier Infra, Qwen3.5-9B LoRA) v1.8.9; open; capability ranked; ranked; estimate 63.4%
  177. 177 GLiClass v4 9B (Knowledgator) v1.8.9; open; capability ranked; ranked; estimate 63.2%
  178. 178 Seb-9B (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 63.1%
  179. 179 wity-1 (Wity, reasoning off) v1.8.9; api; outside capability caps; listed; tariff 63%
  180. 180 Amos 1 medium (Scira) v1.8.9; api; capability ranked; ranked; estimate 62.9%
  181. 181 janus 4B (Icarus AI / cmxu, Qwen3.5-4B + LoRA r64 and pointer decision head) v1.8.9; open; capability ranked; ranked; estimate 62.9%
  182. 182 JevK5 v0.2.0 v1.8.9; open; capability ranked; ranked; estimate 62.8%
  183. 183 Messier One v0.4 (Qwen3.5-4B fine-tune, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 62.4%
  184. 184 Imajev-9B (mohit67890, Qwen3.5-9B LoRA) v1.8.9; open; capability ranked; ranked; estimate 62.4%
  185. 185 Imajev-4B (RTX 5090) v1.8.9; open; capability ranked; ranked; estimate 61.9%
  186. 186 jqv (Qwen3-32B zero-shot) v1.8.9; open; capability ranked; ranked; estimate 61.9%
  187. 187 Open-Jev 9B (Zefan Cai) v1.8.9; open; outside capability caps; ranked; estimate 61.9%
  188. 188 metask-jev-4b v1.8.9; open; capability ranked; ranked; estimate 61.8%
  189. 189 Kahn1 4B (Okura66, Qwen3.5-4B LoRA merge, author's sysone engine) v1.8.9; open; capability ranked; ranked; estimate 61.7%
  190. 190 pplx-decider-flash-v1.2 (Perplexity, Qwen3.5-4B full fine-tune, marker readout, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 61.7%
  191. 191 lev (Interfaze AI, Qwen3.5-4B + LoRA r32, label-token readout + candidate-path head, per-bucket temperatures) v1.8.9; open; capability ranked; ranked; estimate 61.7%
  192. 192 Decision 2.0 Lux 9B (vLLM-SR, Qwen3.5-9B lineage) v1.8.9; open; capability ranked; ranked; estimate 61.6%
  193. 193 JEB-35B-A3B (szybkie.ai, Qwen3.6-35B-A3B LoRA) v1.8.9; open; capability ranked; ranked; estimate 61.6%
  194. 194 Kev-9B v2 (Jared Palmer, Qwen3.5-9B-Base LoRA + pointer head) v1.8.9; open; outside capability caps; listed; unpriced 61.5%
  195. 195 Decisio v0.9.0 on Qwen3.6-35B-A3B (official FP8, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 61.5%
  196. 196 blink-4b (thegovind, Qwen3.5-4B derivative) v1.8.9; open; capability ranked; ranked; estimate 61.4%
  197. 197 Jev-Style-2B v1.8.9; open; outside capability caps; unranked; estimate 61.2%
  198. 198 Hopper v1.8.9; open; capability ranked; ranked; estimate 61.2%
  199. 199 Malkuth-4B (newfull5, Kev post-train) v1.8.9; open; capability ranked; ranked; estimate 61.1%
  200. 200 Qwen3.5-9B Jev-like data-mix v2 v1.8.9; open; capability ranked; ranked; estimate 60.9%
  201. 201 BB-Qwen3.5-4B-LoRA (Babak Barazandeh, LoRA on Qwen3.5-4B-Base, hosted /v1/systemone) v1.8.9; api; outside capability caps; listed; estimated 60.4%
  202. 202 Decision 4B v1.2 (FlyMyJev, Qwen3.5-4B + LoRA) v1.8.9; open; capability ranked; ranked; estimate 60.4%
  203. 203 APUS-OpenJev-v1-35B-A3B (merged bf16 checkpoint-5949 MoE on Qwen3.5-35B-A3B, their own native runtime, full depth, self-hosted) v1.8.9; open; outside capability caps; listed; unpriced 60.4%
  204. 204 jiwo 4b (Qwen3.5-4B backbone with a trained decision head, one forward pass, native probabilities, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 60.2%
  205. 205 PurpleMIST-Flash-1.0 (Olaverse) v1.8.9; open; outside capability caps; listed; unpriced 60.1%
  206. 206 Standard One 8B SH (Standard Thinking) v1.8.9; open; outside capability caps; ranked; estimate 60%
  207. 207 Gemini 3.1 Flash-Lite v1.8.9; api; outside capability caps; ranked; tariff 59.8%
  208. 208 Verdin-4B (Victor Bennett, Qwen3.5-4B + merged LoRA r16 + pointer head) v1.8.9; open; capability ranked; ranked; estimate 59.6%
  209. 209 NVIDIA Nemotron 3.5 Super VL v1.8.9; api; outside capability caps; unranked; tariff 59.6%
  210. 210 Wald 4B v1.8.9; open; capability ranked; ranked; estimate 59.5%
  211. 211 Quyet-1.0-Medium (Chinh Nguyen, Qwen3.5-4B decoder, option-letter logits) v1.8.9; open; capability ranked; ranked; estimate 59.3%
  212. 212 Decision-4B (Eval Engine / Chromia) v1.8.9; open; capability ranked; ranked; estimate 59.1%
  213. 213 reflex 4B (kshetrajna12) v1.8.9; open; outside capability caps; ranked; estimate 58.9%
  214. 214 openjev-sglang (Qwen3.6-35B-A3B on SGLang) v1.8.9; api; outside capability caps; ranked; estimate 58.7%
  215. 215 Standard One 8B (Standard Thinking) v1.8.9; open; outside capability caps; ranked; estimate 58.4%
  216. 216 Clef-omni v1.8.9; open; capability ranked; ranked; estimate 58.3%
  217. 217 Manchego v2.1 v1.8.9; open; capability ranked; ranked; estimate 58.3%
  218. 218 Messier One v0.2 (Qwen3.5-4B fine-tune, one prefill, self-hosted) v1.8.9; open; capability ranked; ranked; estimated 58.1%
  219. 219 Diffusion Jev (DiffusionGemma 26B-A4B on patched SGLang, 48-step diffusion readout, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 58.1%
  220. 220 baselabs joey-1 (2B) v1.8.9; open; outside capability caps; unranked; estimate 57.8%
  221. 221 jev-local (Qwen3.5-9B) v1.8.9; open; capability ranked; ranked; estimate 57.5%
  222. 222 Instinct Dual 4B v1.8.9; api; capability ranked; ranked; tariff 57.2%
  223. 223 Decision 4B v1.1 (FlyMyJev, Qwen3.5-4B + LoRA) v1.8.9; open; capability ranked; ranked; estimate 57.2%
  224. 224 Granite Switch 4.1 8B (revision 4.1.1, decision adapter) v1.8.9; open; outside capability caps; unranked; estimate 57.1%
  225. 225 JEV Qwen3.5-9B Base NVFP4 v1.8.9; open; capability ranked; ranked; estimate 57.1%
  226. 226 iota-decide-4b (Macrocosmos, Clef joint-head 4B, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 57.1%
  227. 227 Uniphore-4B-Decider v1.8.9; open; capability ranked; ranked; estimate 57.1%
  228. 228 Trio-Spark v1.3 (MachineFi) v1.8.9; api; outside capability caps; listed; unpriced 57%
  229. 229 Jebadiah 9B v2 (Frontier Infra, Qwen3.5-9B LoRA) v1.8.9; open; capability ranked; ranked; estimate 57%
  230. 230 CoCo-Decision-4B (corners-ai, LoRA on Qwen3.5-4B, option-label logits of one pass, served by oh-my-jev) v1.8.9; open; capability ranked; ranked; estimate 56.9%
  231. 231 APUS-OpenJev-v1-9B (merged bf16 checkpoint-3000 on Qwen3.5-9B, their own native runtime, full depth, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 56.9%
  232. 232 Decider-4B (roudone, NeoHorse 4B pointer model, author server, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 56.8%
  233. 233 typecastlm (Mikhail Gribov, Qwen3.5-4B computed head) v1.8.9; open; capability ranked; ranked; estimate 56.5%
  234. 234 Bespoke Nimble 9B (Bespoke Labs) v1.8.9; open; capability ranked; ranked; estimate 56.5%
  235. 235 pjev-2b (Kailune AI, frozen MiniCPM5-2B + merged LoRA r64, letter-logit readout, T=1 stub) v1.8.9; open; capability ranked; ranked; estimate 56.5%
  236. 236 TypeCastLM 1.4.0 (Mikhail Gribov, Qwen3.5-4B computed head) v1.8.9; open; capability ranked; ranked; estimate 56.4%
  237. 237 Sors-2B (SakuraYuyuko) v1.8.9; open; outside capability caps; unranked; estimate 56.3%
  238. 238 RYOTIDE-Qwen (csabag, Qwen3.5-4B stock weights) v1.8.9; open; capability ranked; ranked; estimate 56.3%
  239. 239 Amber 4B (MLukauskas) v1.8.9; open; outside capability caps; unranked; estimate 56.1%
  240. 240 Compass 0.2.0 (adimyth, Qwen3.5-4B LoRA) v1.8.9; open; capability ranked; ranked; estimate 56.1%
  241. 241 Standard One 3B SH v4 (Standard Thinking) v1.8.9; open; capability ranked; ranked; estimate 56%
  242. 242 Standard One 3B SH (Standard Thinking) v1.8.9; open; capability ranked; ranked; estimate 55.6%
  243. 243 SemIf, formerly OpenJev (Qwen3.5-4B, TheoLeeCJ) v1.8.9; open; capability ranked; ranked; estimate 55.4%
  244. 244 ArseneLupin v1.1 (IBOU-SEARCH, Qwen3.5-4B LoRA) v1.8.9; open; capability ranked; ranked; estimate 55.1%
  245. 245 local-jev Qwen3.5-4B v1.8.9; open; capability ranked; ranked; estimate 55%
  246. 246 Decision 2B (FlyMy.AI, v59) v1.8.9; open; capability ranked; ranked; estimate 54.8%
  247. 247 Bongard-mini (AgentBull, T5Gemma-2 4B-4B) v1.8.9; open; outside capability caps; listed; unpriced 54.8%
  248. 248 RYOTIDE-Gemma (csabag, Gemma-4-E4B stock weights) v1.8.9; open; outside capability caps; listed; unpriced 54.5%
  249. 249 Jez-large-1.0.0 (pragmaticfrontier, Gemma-4-E4B label-logit decision model, author server, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 54.4%
  250. 250 Noma (Blackdrome AI Labs, Qwen3.5-4B-Base derivative) v1.8.9; open; capability ranked; ranked; estimate 54.3%
  251. 251 spark-s1-4b-v6 (Open Spark Jev, abhishek085) v1.8.9; open; capability ranked; ranked; estimate 54%
  252. 252 Imajev-2B (mohit67890, Qwen3.5-2B LoRA) v1.8.9; open; capability ranked; ranked; estimate 54%
  253. 253 cribl-decision-2-mini (Cribl) v1.8.9; open; outside capability caps; unranked; estimate 53.2%
  254. 254 RYOTIDE-Qwen9 v1.8.9; open; capability ranked; ranked; estimate 52.5%
  255. 255 classify-1 (ScaleDown) v1.8.9; api; capability ranked; ranked; mixed 52.3%
  256. 256 Arbiter-4B (Goddragon, NeoHorse 4B pointer model, author server, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 52.3%
  257. 257 DecisionNode Edge 0.9.2 v1.8.9; open; capability ranked; ranked; estimate 52.1%
  258. 258 system-one-open (Gemma 4 E2B LoRA on an L4) v1.8.9; api; capability ranked; ranked; estimate 51.9%
  259. 259 D1A on Gemma 4 E4B (JohnP1 d1a-e4b) v1.8.9; open; capability ranked; ranked; estimate 51.1%
  260. 260 ClassOne Qwen 3.5 9B (Qwen3.5-9B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 50.5%
  261. 261 kev 8B (research preview) v1.8.9; open; capability ranked; ranked; estimate 50%
  262. 262 Nev-2B (EryriLabs) v1.8.9; open; outside capability caps; remeasure-pending; estimate 49.9%
  263. 263 Liquid AI d1-3B v1.8.9; open; capability ranked; ranked; estimate 49.5%
  264. 264 ZeroEntropy zerank-2 v1.8.9; open; capability ranked; ranked; estimate 49.4%
  265. 265 JADE (Qwen3.8-27B LoRA) v1.8.9; open; capability ranked; ranked; estimate 49.3%
  266. 266 open-alternative-jev (Qwen3.5-4B, IkerMoel) v1.8.9; open; capability ranked; ranked; estimate 49.1%
  267. 267 Qwen3.8 27B (Chutes TEE) v1.8.9; api; outside capability caps; ranked; estimate 48.6%
  268. 268 OpenSourceJev (Qwen3.5-4B Q4_K_M, native llama.cpp) v1.8.9; open; capability ranked; ranked; estimate 48.6%
  269. 269 Open-Jev 2B (Zefan Cai) v1.8.9; open; capability ranked; ranked; estimate 47.7%
  270. 270 EXAONE-4.0-1.2B-JEV v0.3 (unofficial fine-tune of LGAI-EXAONE/EXAONE-4.0-1.2B, native option-label softmax, two reads averaged, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 47.7%
  271. 271 Malkuth-2B (newfull5, Kev post-train) v1.8.9; open; capability ranked; ranked; estimate 47.6%
  272. 272 DecisionNode Edge Mini 0.9.2 v1.8.9; open; capability ranked; ranked; estimate 47.6%
  273. 273 Nemotron Diffusion 8B (pst2154, optimized vLLM) v1.8.9; open; capability ranked; ranked; estimate 47.5%
  274. 274 jiwo 0.8b (Qwen3.5-0.8B backbone with a trained decision head, one forward pass, native probabilities, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 47.2%
  275. 275 decision-machine-1 (milliseconds.ai) v1.8.9; api; capability ranked; ranked; tariff 46.1%
  276. 276 decisor-4b v1.8.9; open; capability ranked; ranked; estimate 45.7%
  277. 277 D1A on Gemma 4 E2B (JohnP1 d1a-e2b) v1.8.9; open; capability ranked; ranked; estimate 45.5%
  278. 278 Qwen3-Reranker-4B v1.8.9; open; capability ranked; ranked; estimate 44.8%
  279. 279 decider-2b (Mapika) v1.8.9; open; capability ranked; ranked; estimate 44.6%
  280. 280 Laya typed-decisions v1.8.9; open; outside capability caps; ranked; estimate 44.2%
  281. 281 BAAI bge-reranker-v2-m3 v1.8.9; open; capability ranked; ranked; estimate 44.1%
  282. 282 kev 4B (research preview) v1.8.9; open; capability ranked; ranked; estimate 44%
  283. 283 Mixedbread mxbai-rerank-base-v2 v1.8.9; open; capability ranked; ranked; estimate 43.9%
  284. 284 lev-350m (Franck Verrot, LFM2.5-350M) v1.8.9; open; capability ranked; ranked; estimate 43.4%
  285. 285 Jez-small-1.0.0 (pragmaticfrontier, Qwen3.5-0.8B label-logit decision model, author server, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 43.2%
  286. 286 Certo v1 (AltSlate Labs) v1.8.9; open; capability ranked; ranked; estimate 43.2%
  287. 287 PurpleMIST-Mini-1.0 (Olaverse) v1.8.9; open; capability ranked; ranked; estimate 43%
  288. 288 Chon-1.0-12B v1.8.9; open; capability ranked; ranked; estimate 43%
  289. 289 smalljev semantic-v9 v1.8.9; open; capability ranked; ranked; estimate 43%
  290. 290 Vega-U (Qwen3.5-4B reader) v1.8.9; open; capability ranked; ranked; estimate 42.6%
  291. 291 jul fast (usejul/jul-decision-minicpm5-2b: MiniCPM5-2B fine-tune with a pointer head, option probabilities from one pass, self-hosted via `jul serve`) v1.8.9; open; capability ranked; ranked; estimate 42.6%
  292. 292 Qwen3.5-0.8B Decision Model (Mourad Ghafiri) v1.8.9; open; outside capability caps; remeasure-pending; estimate 42.6%
  293. 293 Krino-2B v1.8.9; open; capability ranked; ranked; estimate 42.4%
  294. 294 Gutsy 0.8B v0.3 v1.8.9; open; outside capability caps; ranked; estimate 42.3%
  295. 295 Alibaba GTE Reranker ModernBERT-base v1.8.9; open; capability ranked; ranked; estimate 41.7%
  296. 296 Jeetu Decider 2B v1.8.9; open; capability ranked; ranked; estimate 41.7%
  297. 297 watt-flash-0.1 (Zaitgeist Labs, 140M mmBERT-small encoder with an option-marker scorer, one forward pass per request, self-hosted) v1.8.9; open; capability ranked; ranked; mixed 41.6%
  298. 298 Decision Fast (FlyMy.AI, v53a) v1.8.9; open; capability ranked; ranked; estimate 40.6%
  299. 299 openJev Verdict 1.4 v1.8.9; open; capability ranked; ranked; estimate 40.4%
  300. 300 Quyet-1.0-Small-EN (Chinh Nguyen, ModernBERT-base encoder with fixed heads) v1.8.9; open; capability ranked; ranked; estimate 39.5%
  301. 301 Reflex Instinct 0.6B (Atlas AI) v1.8.9; open; outside capability caps; remeasure-pending; estimate 38.6%
  302. 302 ClassOne Gemma 4 E4B (Gemma-4-E4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 38.4%
  303. 303 ClassOne Qwen 3.5 4B (Qwen3.5-4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 38.1%
  304. 304 OpenDecision (ModernBERT-large zero-shot) v1.8.9; open; capability ranked; ranked; estimate 38%
  305. 305 verdict-small (Manavarya09, multilingual-e5-small 118M) v1.8.9; open; capability ranked; ranked; estimate 37.6%
  306. 306 kev 0.6B (research preview) v1.8.9; open; capability ranked; ranked; estimate 36.9%
  307. 307 jevora-9b-v1 (kadirbekovvv / doxanocap, Qwen3.5-9B-Base derivative) v1.8.9; open; outside capability caps; listed; unpriced 36.9%
  308. 308 Vega 4B v1.8.9; open; capability ranked; ranked; estimate 36.6%
  309. 309 Vega-U (Qwen3.5-0.8B reader) v1.8.9; open; outside capability caps; ranked; estimate 36.1%
  310. 310 WaterSheep (Samrat Dutta, ModernBERT-base encoder with calibrated heads, served by the author's own `watersheep --serve`) v1.8.9; open; capability ranked; ranked; estimate 35.5%
  311. 311 open-jev-deberta-v3-large (local CPU) v1.8.9; open; outside capability caps; ranked; estimate 35.4%
  312. 312 Raw Phi-4 mini direct logits v1.8.9; open; capability ranked; ranked; estimate 35.3%
  313. 313 modernbert-jev v2 v1.8.9; open; outside capability caps; remeasure-pending; estimate 34.6%
  314. 314 Fastino GLiNER-2.5-Decide (hosted API) v1.8.9; api; capability ranked; ranked; mixed 34.6%
  315. 315 Raw Qwen3 4B Instruct 2507 direct logits v1.8.9; open; capability ranked; ranked; estimate 34.6%
  316. 316 Vega 0.8B v1.8.9; open; capability ranked; ranked; estimate 33%
  317. 317 kev 0.5B v1.8.9; open; capability ranked; ranked; estimate 32.9%
  318. 318 Liquid AI d1-omni-600M v1.8.9; open; capability ranked; ranked; estimate 32.8%
  319. 319 Laya (Convai Innovations, ModernBERT-large 421M) v1.8.9; open; outside capability caps; ranked; estimate 32.5%
  320. 320 Quyet-1.0-Small (Chinh Nguyen, SEA-LION-ModernBERT-300M encoder with fixed heads) v1.8.9; open; capability ranked; ranked; estimate 32.5%
  321. 321 Quyet-1.0-Tiny (Chinh Nguyen, mmBERT-small encoder (16 layers kept) with fixed heads) v1.8.9; open; capability ranked; ranked; estimate 31.1%
  322. 322 SimpleJev (Qwen3.5-0.8B, CPU) v1.8.9; open; outside capability caps; ranked; estimate 31.1%
  323. 323 TDE general large v0.1 (tdelab, ModernBERT-large encoder) v1.8.9; open; outside capability caps; listed; unpriced 30.9%
  324. 324 Open Jev JSON Canvas (JoshuaSP) v1.8.9; open; capability ranked; ranked; estimate 30.6%
  325. 325 Phocinae-Largha-150M-v1 v1.8.9; open; capability ranked; ranked; estimate 30.2%
  326. 326 Raw Qwen3 8B direct logits v1.8.9; open; capability ranked; ranked; estimate 30.1%
  327. 327 system-one (Qwen3-8B, Sean Goedecke) v1.8.9; open; capability ranked; ranked; estimate 29.7%
  328. 328 JevAct (einptein, jev1-2b-v2) v1.8.9; api; capability ranked; ranked; estimate 28.8%
  329. 329 openJev Verdict (heman10x, ModernBERT-base 151M) v1.8.9; open; capability ranked; ranked; estimate 26.6%
  330. 330 Mirror v1.8.9; open; outside capability caps; ranked; estimate 25.8%
  331. 331 Tacet Sonata (CodePawl, 144M packed encoder on mmBERT-small, option-marker softmax in one pass, in-process via the author's tacet package) v1.8.9; open; capability ranked; ranked; estimate 24.2%
  332. 332 CLM-8B (Contrastive-LM, clm-latest) v1.8.9; open; capability ranked; ranked; estimate 20.1%
  333. 333 ClassOne Qwen 3.5 2B (Qwen3.5-2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 19.2%
  334. 334 Deem 0.8B v1 v1.8.9; open; capability ranked; ranked; estimate 18.1%
  335. 335 Laya multilingual v1.8.9; open; capability ranked; ranked; estimate 17%
  336. 336 ClassOne Gemma 4 E2B (Gemma-4-E2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) v1.8.9; open; capability ranked; ranked; estimate 15.2%
  337. 337 Julia-1 (Supersonic Labs) v1.8.9; open; capability ranked; ranked; estimate 9.85%
  338. 338 Raw Qwen3 1.7B direct logits v1.8.9; open; capability ranked; ranked; estimate 8.32%
  339. 339 Raw Qwen3 0.6B direct logits v1.8.9; open; capability ranked; ranked; estimate 8.25%

Source: Official JevBench leaderboard

#EntryScoreSettingDateSource
1 GPT-6 Luna 98.4% v1.8.9; api; outside capability caps; ranked; tariff Sep 2026 Official leaderboard
2 GPT-5.6 Luna 93.6% v1.8.9; api; outside capability caps; ranked; tariff Jul 2026 Official leaderboard
3 Torchcast Decision 31B v1.0 91.8% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
4 MetaDecision-31B-Rain 90.7% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
5 DeepSeek V4.1 Flash 89.6% v1.8.9; api; outside capability caps; ranked; tariff Sep 2026 Official leaderboard
6 Karpo-Decision-31B 89.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
7 WeMM-JEV (Rover00) 88.7% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
8 Jeff-1.1-Large (jgeuter) 88.2% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
9 VisionJEV 1.0.0 86.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
10 kettei-fat (Sqwish) 86.3% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
11 Piccolo 27B (agentscope-ai) 86% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
12 Super Decider (thinking) 85.8% v1.8.9; api; capability ranked; ranked; listed Official leaderboard
13 baselabs mammoth-1 (27B) 85.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
14 KnowLine 31B-Gen1 85.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
15 Fastino GLiDE 1.2 (open weights, self-hosted) 84.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
16 Super Decider 84.9% v1.8.9; api; capability ranked; ranked; listed Official leaderboard
17 Fastino GLiDE (hosted API, 22 Oct 2026) 84.5% v1.8.9; api; capability ranked; ranked; mixed Official leaderboard
18 H2O-Lightning-31B v1.0.7 (H2O.ai, Gemma-4-31B-it fine-tune, vLLM + h2oai overlay, one-engine FP8 profile, self-hosted) 84.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
19 Drex Max (Nace AI) 84.2% v1.8.9; api; capability ranked; ranked; listed Official leaderboard
20 decide-1.1 (decide.sh, API) 84% v1.8.9; api; outside capability caps; unranked; tariff Official leaderboard
21 oenogarum (plantwaterer; commercially restricted) 83.9% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
22 allec-oavg (plantwaterer; commercially restricted) 83.2% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
23 Perplexity Decider v1.1 27B (Qwen3.8-27B, noncausal decision readout) 82.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
24 vLLM-SR d3-turbo 82.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
25 StartLux-Decision v1.1 27B Instant (StartLux Labs, Qwen3.8-27B fine-tune) 82.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
26 Decision-King v1.1 (xsdfasfgsa) 82.3% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
27 vLLM-SR d3 82.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
28 basal-omni-1.0-ultra (Remek Kinas, Gemma-4-31B-it + merged LoRA, typed-decision readout, self-hosted) 82.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
29 DOPEST 1.0 31B 82% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
30 SubQ Snap 1.1 (SubQ) 81.9% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
31 Jeff-1.1-Medium (jgeuter) 81.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
32 Arcbright Elf 1 31B 81.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
33 Quyet-1.0-Large (Chinh Nguyen, Gemma-4-31B decoder, option-letter logits) 81.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
34 Jeffan-31b-1710 81.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
35 Jeff-1.0-Large 81.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
36 SubQ Snap (preview, SubQ) 81.5% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
37 AJev lm6q (Gemma 4 31B) 81.4% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
38 Nimble-v4-max (open weights, self-hosted) 81.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
39 Sieve-31B-v2 81.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
40 Turion Jev 31B 80.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
41 pplx-decider-v1.2 (Perplexity, Qwen3.8-27B full fine-tune, marker readout, self-hosted) 80.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
42 d1-ultra (Liquid AI) 80.5% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
43 torchcast-decision-27b (Torchcast AI, Qwen3.8-27B fine-tune) 80.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
44 Arcbright Elf 1 12B (r2) 79.9% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
45 Lychee-27B 79.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
46 Microsoft-Decision-1 (updated weights, Oct 2026) 79.8% v1.8.9; api; outside capability caps; unranked; tariff Official leaderboard
47 Bespoke Nimble-v4-max (Bespoke Labs) 79.7% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
48 Jeff-1.1-Qwen-4B (jgeuter) 79.6% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
49 metask-jev-rain-31B-trained service v3 79.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
50 decisio v0.8.0 on gemma-4-31B-it (frozen, FP8 on load, one prefill per state, self-hosted) 79.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
51 Vera Large 79.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
52 Kev 27B (Kev 1.0, Qwen3.8-27B) 78.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
53 metask-jev-rain-31B-trained 78.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
54 Myles-G 31B (Launchables) 78.8% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
55 StartLux-Decision v1.1 27B AutoThink (StartLux Labs, Qwen3.8-27B fine-tune, gate 0.6) 78.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
56 Decisio 12B tuned commit-think (variant of the Decisio family) 78.7% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
57 Sage 1.3.0 (Levanto Labs) 78.6% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
58 SPX-CD Pro (SurdAI, hosted /v1/systemone, Oct-4 checkpoint) 78.5% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
59 metask-jev-rain-12B 78.3% v1.8.9; open; outside capability caps; wrapper; estimate Official leaderboard
60 classifier.dev (fast tier) 78.3% v1.8.9; api; outside capability caps; wrapper; estimate Official leaderboard
61 Amos 1 large (Scira) 78% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
62 Bobcat 1.3 (Gemma 4 31B) 77.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
63 deck-31B (krishna765, frozen Gemma-4-31B-it, TorchAO FP8 dynamic) 77.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
64 SD-JEV-4B-v12.1.0 (Kego, author-hosted API) 77.5% v1.8.9; api; outside capability caps; unranked; estimate Official leaderboard
65 Torchcast Decision 4B (v1.3) 77.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
66 KnowLine 26B-A4B-Gen2.5 77.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
67 Eikos-27B (caiovicentino1, Qwen3.8-27B) 77.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
68 Jev 1.13.0 (TypeSafe AI) 77.1% v1.8.9; reference; capability ranked; ranked; tariff Official leaderboard
69 wity-1 (Wity, reasoning always) 77.1% v1.8.9; api; outside capability caps; listed; tariff Official leaderboard
70 H2O-Lightning-4B v1.4.1 77% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
71 Fastino GLiDE 77% v1.8.9; api; outside capability caps; ranked; tariff Official leaderboard
72 wity-1 (Wity, reasoning auto) 76.9% v1.8.9; api; outside capability caps; ranked; tariff Official leaderboard
73 metask-jev-rain-12B-trained 76.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
74 Aleria Judge 1.1.1 (ALERIA-ML, inference technique on stock Gemma 4 31B, author server, self-hosted) 76.9% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
75 Wald 26B-A4B (Gemma 4 26B-A4B) 76.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
76 DecisionNode 76.8% v1.8.9; api; outside capability caps; unranked; unpriced Official leaderboard
77 KnowLine 35B-A3B-Gen1.5 76.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
78 Piccolo 9B (agentscope-ai) 76.6% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
79 cribl-decision-1.0-27B (Cribl) 76.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
80 René-1 31B FP8 (salfatigroup, Gemma 4 31B full fine-tune, one-pass option readout) 76% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
81 OpenJev (thinking, BF16) 75.8% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
82 Decisio 12B tuned commit (variant of the Decisio family) 75.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
83 Clef (Cloudflare, Qwen3.8-27B post-train with a joint schema head, multimodal, measured on text) 75.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
84 Jev 1.14.0 (TypeSafe AI) 75% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
85 H2O-Lightning-4B v1.1 (H2O.ai, Qwen3.5-4B fine-tune, stock vLLM + open shim) 75% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
86 SimpleJev Qwen3.8-27B (self-hosted, v1.6 pool) 74.9% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
87 Umpire 35B-A3B 74.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
88 z3-mini-soup-preview (Zils AI) 74.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
89 d1 (Liquid AI) 74.4% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
90 SPX-CD Flash (SurdAI, hosted /v1/systemone, Oct-4 checkpoint) 74.3% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
91 iota-decide-4b-v0.4 (Macrocosmos, letter-readout 4B, self-hosted) 74% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
92 JEV-27B (AutoTrust, Qwen3.8-27B) 73.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
93 Nagi-ENORMOUS 27B (nagisanzenin, Qwen3.8-27B LoRA) 73.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
94 Mercury Decide (Inception; System One decisions API, served free on OpenRouter as inception/mercury-decide:free) 73.6% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
95 Celeris-1 Decision (Celeris) 73.6% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
96 Piccolo 4B (agentscope-ai) 73.6% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
97 OpenAI Decisions (gpt-6-luna) 73.5% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
98 Surogate Rune 26B-A4B v3 (v1.6 pool) 73.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
99 RSI-Jev v6.1-VL 27B 73.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
100 Sage 1.4 (Levanto Labs) 73.3% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
101 kettei-chub (Sqwish) 73.1% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
102 Bobcat Flash 1.2 (Gemma-4-26B-A4B-it + two merged rank-64 LoRA adapters, typed-decision readout at the first answer position, FP8, self-hosted) 73% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
103 Autoloops – Gemma 4 31B IT 72.7% v1.8.9; api; outside capability caps; ranked; tariff Official leaderboard
104 Torchcast Decision 4B (v1.2) 72.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
105 swanOne (blockbrain, Qwen3.8-Flash-Next NVFP4) 72.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
106 FrontiersMind Lumma-fev-27B 72.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
107 Microsoft-Decision-1 (Azure Foundry) — A2 re-run 72% v1.8.9; api; outside capability caps; preliminary; tariff Official leaderboard
108 torchcast-decision-12b (Torchcast AI, Gemma-4-12B fine-tune, option-letter logprob readout) 71.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
109 AutoJev-27B (denis-pplx, Qwen3.8-27B) 71.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
110 SimpleJev Qwen3.8-27B 71.6% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
111 decider-12b v2 (Mapika) 71.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
112 Decision 2.0 Vega 27B (vLLM-SR, Qwen3.8-27B) 71.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
113 Jev-Omni (akhilaaa3, Gemma-4-12B merged) 71.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
114 Winnow-12B Q8 71.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
115 Xor 26B-A4B (Juspay, Gemma-4-26B-A4B, bf16) 71.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
116 decider-12b v1, stock Gemma-4-12B-it (Mapika) 71% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
117 Cygnet (blockbrain, frozen Gemma-4-12B-it) 70.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
118 Microsoft-Decision-1 (Azure Foundry) 70.8% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
119 decisio v0.8.0 on gemma-4-12B-it (frozen, one forward pass, self-hosted) 70.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
120 JPT-35B-A3B (Qwen3.5-35B-A3B fine-tune) 70.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
121 Myles-G 12B (Launchables) 70.7% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
122 Instinct (ZooWork, Qwen3.8-27B) 70.5% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
123 decider chat on Gemma-4-31B-it (Mapika, frozen base, inference technique) 70.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
124 baselabs elephant-1 (9B) 70.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
125 Jebadiah 27B (Frontier Infra, Qwen3.8-27B LoRA) 70.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
126 GEV-26B-Decide (AutoTrust, Gemma-4-26B-A4B + LoRA + head) 70.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
127 Chon-1.0-12B-Fast 70.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
128 Blink v0.3 26B-A4B 70% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
129 decisio v0.9.0 on gemma-4-12B-it (frozen, one prefill per question, self-hosted) 70% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
130 NInfer Qwen3.8-Flash-Next mixed 69.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
131 attest-31B-v1.0 69.5% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
132 Respan Span-02 69.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
133 Nemotron Deciffusion 8B v2 (pst2154) 69.3% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
134 Philly Decider v2 (PhillyLabs) 69.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
135 Wald 4B v2.1 69.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
136 JevOne 68.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
137 vLLM-SR d3-mini 68.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
138 Torchcast Decision 4B (v1.1) 68.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
139 iota-decide-4b-v0.3 (Macrocosmos, letter-readout 4B, self-hosted) 68.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
140 Xevn-4B 68.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
141 Chon-1.0-E4B 68.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
142 Strike-1 68.3% v1.8.9; api; outside capability caps; unranked; estimate Official leaderboard
143 DecisionNode Flash 68.3% v1.8.9; api; outside capability caps; unranked; unpriced Official leaderboard
144 SimpleJev Qwen3.6-35B-A3B 68.3% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
145 Hopper 12B trained (gemma-4-12B-it frozen + LoRA r32 unmerged, one forward pass, self-hosted) 68.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
146 SPX-CD-Omni (SurdAI, google/gemma-4-12B-it + LoRA r32, one forward pass, self-hosted) 68.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
147 NInfer Qwen3.8-27B NVFP4 68% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
148 Open-Jev 27B v1.1 (Zefan Cai, LoRA + decision head on Qwen3.8-27B) 68% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
149 Aplomb 1 (5.3B decision model on Qwen3.5-4B, trained readout head, self-hosted) 67.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
150 Bespoke Nimble 9B v3 (Qwen3.5-9B LoRA) 66.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
151 Amazon Strands Decider 26B-A4B 66.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
152 LitJev (Qwen3.8-27B) 66.6% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
153 Saif 4B (Gyr Technology) 66.4% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
154 Tern-35B-A3B (matthewagi, Qwen3.6 35B-A3B MoE decision model, author server, self-hosted) 66.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
155 KnowLine 4B-Gen5.5 66.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
156 decider-35b-a3b (Mapika) 66.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
157 Bev / Bonsai 27B 66.1% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
158 Metask rain 4B 66% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
159 cribl-decision-1.0-9B (Cribl) 65.9% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
160 shisa-de-1 (shisa-ai, Gemma-4-26B-A4B) 65.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
161 Plumb-4B (crh225, JevK5 v0.2 + LoRA) 65.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
162 cribl-decision-2 (Cribl) 65.1% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
163 Arcbright Elf 1 12B 65% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
164 decider-4b v2 (Mapika) 64.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
165 Seb-9B-v2 (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted) 64.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
166 Granite Switch 4.1 30B (revision 4.1.1, decision adapter) 64.5% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
167 HS1 0.1.2 (Hyland) 64.3% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
168 JevK5 v0.3 (4B) 64.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
169 Nagi-HUGE 12B (nagisanzenin, Gemma-4-12B LoRA) 64% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
170 Bespoke Nimble-v4 (Bespoke Labs) 64% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
171 Jeffan-4B-2010 64% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
172 Clef-Flash (Cloudflare, Qwen3.5-9B post-train with a joint schema head, multimodal, measured on text) 63.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
173 JPT-4B (kirp, Qwen3.5-4B LoRA) 63.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
174 Vansa-3.4 (Vansa, hosted System One API) 63.7% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
175 deck-4B v1.0 (krishna765, JevK5 + LoRA, FP8 weight-only) 63.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
176 Jebadiah 9B v2.1 (Frontier Infra, Qwen3.5-9B LoRA) 63.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
177 GLiClass v4 9B (Knowledgator) 63.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
178 Seb-9B (ironbcc, Qwen3.5-9B fine-tune, one forward pass, stock vLLM, self-hosted) 63.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
179 wity-1 (Wity, reasoning off) 63% v1.8.9; api; outside capability caps; listed; tariff Official leaderboard
180 Amos 1 medium (Scira) 62.9% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
181 janus 4B (Icarus AI / cmxu, Qwen3.5-4B + LoRA r64 and pointer decision head) 62.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
182 JevK5 v0.2.0 62.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
183 Messier One v0.4 (Qwen3.5-4B fine-tune, self-hosted) 62.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
184 Imajev-9B (mohit67890, Qwen3.5-9B LoRA) 62.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
185 Imajev-4B (RTX 5090) 61.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
186 jqv (Qwen3-32B zero-shot) 61.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
187 Open-Jev 9B (Zefan Cai) 61.9% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
188 metask-jev-4b 61.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
189 Kahn1 4B (Okura66, Qwen3.5-4B LoRA merge, author's sysone engine) 61.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
190 pplx-decider-flash-v1.2 (Perplexity, Qwen3.5-4B full fine-tune, marker readout, self-hosted) 61.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
191 lev (Interfaze AI, Qwen3.5-4B + LoRA r32, label-token readout + candidate-path head, per-bucket temperatures) 61.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
192 Decision 2.0 Lux 9B (vLLM-SR, Qwen3.5-9B lineage) 61.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
193 JEB-35B-A3B (szybkie.ai, Qwen3.6-35B-A3B LoRA) 61.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
194 Kev-9B v2 (Jared Palmer, Qwen3.5-9B-Base LoRA + pointer head) 61.5% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
195 Decisio v0.9.0 on Qwen3.6-35B-A3B (official FP8, self-hosted) 61.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
196 blink-4b (thegovind, Qwen3.5-4B derivative) 61.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
197 Jev-Style-2B 61.2% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
198 Hopper 61.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
199 Malkuth-4B (newfull5, Kev post-train) 61.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
200 Qwen3.5-9B Jev-like data-mix v2 60.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
201 BB-Qwen3.5-4B-LoRA (Babak Barazandeh, LoRA on Qwen3.5-4B-Base, hosted /v1/systemone) 60.4% v1.8.9; api; outside capability caps; listed; estimated Official leaderboard
202 Decision 4B v1.2 (FlyMyJev, Qwen3.5-4B + LoRA) 60.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
203 APUS-OpenJev-v1-35B-A3B (merged bf16 checkpoint-5949 MoE on Qwen3.5-35B-A3B, their own native runtime, full depth, self-hosted) 60.4% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
204 jiwo 4b (Qwen3.5-4B backbone with a trained decision head, one forward pass, native probabilities, self-hosted) 60.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
205 PurpleMIST-Flash-1.0 (Olaverse) 60.1% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
206 Standard One 8B SH (Standard Thinking) 60% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
207 Gemini 3.1 Flash-Lite 59.8% v1.8.9; api; outside capability caps; ranked; tariff Mar 2026 Official leaderboard
208 Verdin-4B (Victor Bennett, Qwen3.5-4B + merged LoRA r16 + pointer head) 59.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
209 NVIDIA Nemotron 3.5 Super VL 59.6% v1.8.9; api; outside capability caps; unranked; tariff Official leaderboard
210 Wald 4B 59.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
211 Quyet-1.0-Medium (Chinh Nguyen, Qwen3.5-4B decoder, option-letter logits) 59.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
212 Decision-4B (Eval Engine / Chromia) 59.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
213 reflex 4B (kshetrajna12) 58.9% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
214 openjev-sglang (Qwen3.6-35B-A3B on SGLang) 58.7% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
215 Standard One 8B (Standard Thinking) 58.4% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
216 Clef-omni 58.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
217 Manchego v2.1 58.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
218 Messier One v0.2 (Qwen3.5-4B fine-tune, one prefill, self-hosted) 58.1% v1.8.9; open; capability ranked; ranked; estimated Official leaderboard
219 Diffusion Jev (DiffusionGemma 26B-A4B on patched SGLang, 48-step diffusion readout, self-hosted) 58.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
220 baselabs joey-1 (2B) 57.8% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
221 jev-local (Qwen3.5-9B) 57.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
222 Instinct Dual 4B 57.2% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
223 Decision 4B v1.1 (FlyMyJev, Qwen3.5-4B + LoRA) 57.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
224 Granite Switch 4.1 8B (revision 4.1.1, decision adapter) 57.1% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
225 JEV Qwen3.5-9B Base NVFP4 57.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
226 iota-decide-4b (Macrocosmos, Clef joint-head 4B, self-hosted) 57.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
227 Uniphore-4B-Decider 57.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
228 Trio-Spark v1.3 (MachineFi) 57% v1.8.9; api; outside capability caps; listed; unpriced Official leaderboard
229 Jebadiah 9B v2 (Frontier Infra, Qwen3.5-9B LoRA) 57% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
230 CoCo-Decision-4B (corners-ai, LoRA on Qwen3.5-4B, option-label logits of one pass, served by oh-my-jev) 56.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
231 APUS-OpenJev-v1-9B (merged bf16 checkpoint-3000 on Qwen3.5-9B, their own native runtime, full depth, self-hosted) 56.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
232 Decider-4B (roudone, NeoHorse 4B pointer model, author server, self-hosted) 56.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
233 typecastlm (Mikhail Gribov, Qwen3.5-4B computed head) 56.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
234 Bespoke Nimble 9B (Bespoke Labs) 56.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
235 pjev-2b (Kailune AI, frozen MiniCPM5-2B + merged LoRA r64, letter-logit readout, T=1 stub) 56.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
236 TypeCastLM 1.4.0 (Mikhail Gribov, Qwen3.5-4B computed head) 56.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
237 Sors-2B (SakuraYuyuko) 56.3% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
238 RYOTIDE-Qwen (csabag, Qwen3.5-4B stock weights) 56.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
239 Amber 4B (MLukauskas) 56.1% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
240 Compass 0.2.0 (adimyth, Qwen3.5-4B LoRA) 56.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
241 Standard One 3B SH v4 (Standard Thinking) 56% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
242 Standard One 3B SH (Standard Thinking) 55.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
243 SemIf, formerly OpenJev (Qwen3.5-4B, TheoLeeCJ) 55.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
244 ArseneLupin v1.1 (IBOU-SEARCH, Qwen3.5-4B LoRA) 55.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
245 local-jev Qwen3.5-4B 55% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
246 Decision 2B (FlyMy.AI, v59) 54.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
247 Bongard-mini (AgentBull, T5Gemma-2 4B-4B) 54.8% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
248 RYOTIDE-Gemma (csabag, Gemma-4-E4B stock weights) 54.5% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
249 Jez-large-1.0.0 (pragmaticfrontier, Gemma-4-E4B label-logit decision model, author server, self-hosted) 54.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
250 Noma (Blackdrome AI Labs, Qwen3.5-4B-Base derivative) 54.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
251 spark-s1-4b-v6 (Open Spark Jev, abhishek085) 54% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
252 Imajev-2B (mohit67890, Qwen3.5-2B LoRA) 54% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
253 cribl-decision-2-mini (Cribl) 53.2% v1.8.9; open; outside capability caps; unranked; estimate Official leaderboard
254 RYOTIDE-Qwen9 52.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
255 classify-1 (ScaleDown) 52.3% v1.8.9; api; capability ranked; ranked; mixed Official leaderboard
256 Arbiter-4B (Goddragon, NeoHorse 4B pointer model, author server, self-hosted) 52.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
257 DecisionNode Edge 0.9.2 52.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
258 system-one-open (Gemma 4 E2B LoRA on an L4) 51.9% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
259 D1A on Gemma 4 E4B (JohnP1 d1a-e4b) 51.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
260 ClassOne Qwen 3.5 9B (Qwen3.5-9B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) 50.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
261 kev 8B (research preview) 50% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
262 Nev-2B (EryriLabs) 49.9% v1.8.9; open; outside capability caps; remeasure-pending; estimate Official leaderboard
263 Liquid AI d1-3B 49.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
264 ZeroEntropy zerank-2 49.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
265 JADE (Qwen3.8-27B LoRA) 49.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
266 open-alternative-jev (Qwen3.5-4B, IkerMoel) 49.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
267 Qwen3.8 27B (Chutes TEE) 48.6% v1.8.9; api; outside capability caps; ranked; estimate Official leaderboard
268 OpenSourceJev (Qwen3.5-4B Q4_K_M, native llama.cpp) 48.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
269 Open-Jev 2B (Zefan Cai) 47.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
270 EXAONE-4.0-1.2B-JEV v0.3 (unofficial fine-tune of LGAI-EXAONE/EXAONE-4.0-1.2B, native option-label softmax, two reads averaged, self-hosted) 47.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
271 Malkuth-2B (newfull5, Kev post-train) 47.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
272 DecisionNode Edge Mini 0.9.2 47.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
273 Nemotron Diffusion 8B (pst2154, optimized vLLM) 47.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
274 jiwo 0.8b (Qwen3.5-0.8B backbone with a trained decision head, one forward pass, native probabilities, self-hosted) 47.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
275 decision-machine-1 (milliseconds.ai) 46.1% v1.8.9; api; capability ranked; ranked; tariff Official leaderboard
276 decisor-4b 45.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
277 D1A on Gemma 4 E2B (JohnP1 d1a-e2b) 45.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
278 Qwen3-Reranker-4B 44.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
279 decider-2b (Mapika) 44.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
280 Laya typed-decisions 44.2% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
281 BAAI bge-reranker-v2-m3 44.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
282 kev 4B (research preview) 44% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
283 Mixedbread mxbai-rerank-base-v2 43.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
284 lev-350m (Franck Verrot, LFM2.5-350M) 43.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
285 Jez-small-1.0.0 (pragmaticfrontier, Qwen3.5-0.8B label-logit decision model, author server, self-hosted) 43.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
286 Certo v1 (AltSlate Labs) 43.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
287 PurpleMIST-Mini-1.0 (Olaverse) 43% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
288 Chon-1.0-12B 43% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
289 smalljev semantic-v9 43% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
290 Vega-U (Qwen3.5-4B reader) 42.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
291 jul fast (usejul/jul-decision-minicpm5-2b: MiniCPM5-2B fine-tune with a pointer head, option probabilities from one pass, self-hosted via `jul serve`) 42.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
292 Qwen3.5-0.8B Decision Model (Mourad Ghafiri) 42.6% v1.8.9; open; outside capability caps; remeasure-pending; estimate Official leaderboard
293 Krino-2B 42.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
294 Gutsy 0.8B v0.3 42.3% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
295 Alibaba GTE Reranker ModernBERT-base 41.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
296 Jeetu Decider 2B 41.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
297 watt-flash-0.1 (Zaitgeist Labs, 140M mmBERT-small encoder with an option-marker scorer, one forward pass per request, self-hosted) 41.6% v1.8.9; open; capability ranked; ranked; mixed Official leaderboard
298 Decision Fast (FlyMy.AI, v53a) 40.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
299 openJev Verdict 1.4 40.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
300 Quyet-1.0-Small-EN (Chinh Nguyen, ModernBERT-base encoder with fixed heads) 39.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
301 Reflex Instinct 0.6B (Atlas AI) 38.6% v1.8.9; open; outside capability caps; remeasure-pending; estimate Official leaderboard
302 ClassOne Gemma 4 E4B (Gemma-4-E4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) 38.4% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
303 ClassOne Qwen 3.5 4B (Qwen3.5-4B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) 38.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
304 OpenDecision (ModernBERT-large zero-shot) 38% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
305 verdict-small (Manavarya09, multilingual-e5-small 118M) 37.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
306 kev 0.6B (research preview) 36.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
307 jevora-9b-v1 (kadirbekovvv / doxanocap, Qwen3.5-9B-Base derivative) 36.9% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
308 Vega 4B 36.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
309 Vega-U (Qwen3.5-0.8B reader) 36.1% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
310 WaterSheep (Samrat Dutta, ModernBERT-base encoder with calibrated heads, served by the author's own `watersheep --serve`) 35.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
311 open-jev-deberta-v3-large (local CPU) 35.4% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
312 Raw Phi-4 mini direct logits 35.3% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
313 modernbert-jev v2 34.6% v1.8.9; open; outside capability caps; remeasure-pending; estimate Official leaderboard
314 Fastino GLiNER-2.5-Decide (hosted API) 34.6% v1.8.9; api; capability ranked; ranked; mixed Official leaderboard
315 Raw Qwen3 4B Instruct 2507 direct logits 34.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
316 Vega 0.8B 33% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
317 kev 0.5B 32.9% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
318 Liquid AI d1-omni-600M 32.8% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
319 Laya (Convai Innovations, ModernBERT-large 421M) 32.5% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
320 Quyet-1.0-Small (Chinh Nguyen, SEA-LION-ModernBERT-300M encoder with fixed heads) 32.5% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
321 Quyet-1.0-Tiny (Chinh Nguyen, mmBERT-small encoder (16 layers kept) with fixed heads) 31.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
322 SimpleJev (Qwen3.5-0.8B, CPU) 31.1% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
323 TDE general large v0.1 (tdelab, ModernBERT-large encoder) 30.9% v1.8.9; open; outside capability caps; listed; unpriced Official leaderboard
324 Open Jev JSON Canvas (JoshuaSP) 30.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
325 Phocinae-Largha-150M-v1 30.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
326 Raw Qwen3 8B direct logits 30.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
327 system-one (Qwen3-8B, Sean Goedecke) 29.7% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
328 JevAct (einptein, jev1-2b-v2) 28.8% v1.8.9; api; capability ranked; ranked; estimate Official leaderboard
329 openJev Verdict (heman10x, ModernBERT-base 151M) 26.6% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
330 Mirror 25.8% v1.8.9; open; outside capability caps; ranked; estimate Official leaderboard
331 Tacet Sonata (CodePawl, 144M packed encoder on mmBERT-small, option-marker softmax in one pass, in-process via the author's tacet package) 24.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
332 CLM-8B (Contrastive-LM, clm-latest) 20.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
333 ClassOne Qwen 3.5 2B (Qwen3.5-2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) 19.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
334 Deem 0.8B v1 18.1% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
335 Laya multilingual 17% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
336 ClassOne Gemma 4 E2B (Gemma-4-E2B backbone with ClassOne decision heads, schema-constrained single forward pass, self-hosted) 15.2% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
337 Julia-1 (Supersonic Labs) 9.85% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
338 Raw Qwen3 1.7B direct logits 8.32% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard
339 Raw Qwen3 0.6B direct logits 8.25% v1.8.9; open; capability ranked; ranked; estimate Official leaderboard

2 models have held number one on the official leaderboard since its release in September 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Passed 85% in Sep 2026, by GPT-5.6 Luna
GPT-6 Luna 98.4%

Source: Official JevBench leaderboard

GPT-5.6 Luna on the official leaderboard passed 85% in July 2026

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6 Luna GPT-5.6 Luna Gemini 3.1 Flash-Lite 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6 Luna Gemini 3.1 Flash-Lite

Source: Official JevBench leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Other

All 43 in Other
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Gemini 4 Carbon 59.0
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. sonar 42%
  • Anthropic ECI (AECI) Only in system cards Index that combines a basket of capability benchmarks into one capability score to track trends across Anthropic models. Claude Opus 5.5 169.4
  • SPEED-Bench Only in system cards Measures speculative decoding across varied prompts and domains. Reports the average acceptance length of draft tokens, so higher means faster generation at the same output. Nemotron 3 Ultra 5.15