Benchmark Almanac

Evalatro v2

GLM-5.3-Flash-NVFP4-Spark-RTX-PRO-Preview 46.2 tops its leaderboard MiMo-V2.5-Pro 46.1 leads the models we track
Not in model cards

Models play Balatro from the game state and try to clear Ante 12. Illegal moves reduce the score for progress through the game.

The public leaderboard has 63 scored configurations across 468 games. The maker flagged a response parsing problem that may have affected some runs.

Made by
alesha-pro
Released
June 2026
Kind
Video games
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Agentic
Environment
Balatro
Items from
Generated
Run by
Anyone
In views
Games, Agents
Strongest model tested
GPT-5.6 Sol #14 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
62
Entries on its own leaderboard, 28 models we track

GLM-5.3-Flash-NVFP4-Spark-RTX-PRO-Preview tops the leaderboard with 46.2 MiMo-V2.5-Pro leads with 46.1 among models released by August 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 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 best model we track is MiMo-V2.5-Pro, at 46.1. This view shows only models we track, each at its best.

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

62 entries. Scroll sideways for the rest.

  1. 46.2 GLM-5.3-Flash-NVFP4-Spark-RTX-PRO-Preview
  2. 46.1 DeepSeek-V4.1-Flash-1017-RTX-PRO-Max-Thinking
  3. 46.1 MiMo-V2.5-Pro
  4. 45.5 Qwen3.7-Max
  5. 43.5 deepseek-ai-DeepSeek-V4-Flash-0731-RTX-PRO-Preview
  6. 43.2 DeepSeek-V4-Pro
  7. 41.8 Gemini 3.1 Pro
  8. 41.6 DeepSeek-V4.1-Flash-1017-RTX-PRO-Preview
  9. 41.0 deepseek-ai-DeepSeek-V4-Flash-0731-GB10-Preview
  10. 39.6 Grok 4.5
  11. 39.2 Gemini 3.5 Flash
  12. 39.1 Muse Spark 1.1
  13. 39.0 Gemma 4 31B IT
  14. 36.9 unsloth-Kimi-K3-GGUF-UD-IQ2_XXS-Reasoning-High
  15. 36.3 Inkling
  16. 35.9 DeepSeek-V4-Flash
  17. 35.4 deepseek-ai-DeepSeek-V4-Flash-RTX-PRO
  18. 35.3 unsloth-GLM-5.2-UD-IQ4_XS
  19. 35.0 Qwen3.7-Plus
  20. 31.2 qwen3.6-27b-mtp@q6_k
  21. 29.9 deepseek-ai-DeepSeek-V4-Flash
  22. 29.7 Kimi K2.7 Code
  23. 29.6 GLM-5.3-Flash-NVFP4-GB10-Preview
  24. 26.4 nvidia/nemotron-3-ultra-550b-a55b
  25. 25.2 GLM-5.3-Flash-NVFP4-RTX-PRO-Preview
  26. 23.8 GLM-5.2
  27. 23.3 NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4
  28. 21.7 unsloth-Qwen3.6-27B-MTP-GGUF-Q8_0
  29. 19.6 Nex-N2-Pro
  30. 19.3 qwen3.6-27b-int8-havenoammo
  31. 18.8 Qwen3.8-27B-Q4_K_M
  32. 18.7 Step 3.7 Flash
  33. 18.3 google/gemma-4-26b-a4b-qat
  34. 18.1 qwen3.6-27b-autoround
  35. 17.5 Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M
  36. 17.2 Qwen-Qwen3.8-27B-FP8-RTX-PRO-Preview
  37. 15.0 MiniMax-M3
  38. 13.3 Qwen-AgentWorld-35B-A3B-UD-Q4_K_M
  39. 13.3 gpt-oss-120b
  40. 11.3 unsloth-Qwen3.6-35B-A3B-MTP-GGUF-Q8_K_XL
  41. 11.0 GLM-4.6V-Flash
  42. 10.3 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16
  43. 9.80 unsloth-GLM-5.2-UD-Q6-K-XL
  44. 8.50 Qwen 3.6 35B-A3B
  45. 8.50 Tencent Hy3
  46. 7.50 NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Q4_K_M
  47. 7.00 nvidia/GLM-5.1-NVFP4
  48. 6.50 Qwen 3.8 27B
  49. 5.50 cyankiwi/MiniMax-M2.7-AWQ-4bit
  50. 5.50 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16.gguf
  51. 4.80 gpt-oss-20b
  52. 4.80 MiniMax-M2.7
  53. 4.70 unsloth-GLM-4.6V-UD-Q4_K_XL
  54. 4.20 LFM2.5-8B-A1B
  55. 3.30 /data/models/safetensors/MiniMaxAI/MiniMax-M2.7
  56. 2.50 Ministral 3 3B Instruct
  57. 2.40 GPT-5.6 Terra
  58. 2.20 Gemma 4 26B A4B
  59. 2.10 openai/gpt-5.6-terra-pro
  60. 0.70 lfm2.5-350m
  61. 0.30 Qwen3.5-4B
  62. 0.10 GPT-5.6 Sol
  1. 1 GLM-5.3-Flash-NVFP4-Spark-RTX-PRO-Preview 46.2
  2. 2 DeepSeek-V4.1-Flash-1017-RTX-PRO-Max-Thinking 46.1
  3. 3 MiMo-V2.5-Pro 46.1
  4. 4 Qwen3.7-Max 45.5
  5. 5 deepseek-ai-DeepSeek-V4-Flash-0731-RTX-PRO-Preview 43.5
  6. 6 DeepSeek-V4-Pro 43.2
  7. 7 Gemini 3.1 Pro 41.8
  8. 8 DeepSeek-V4.1-Flash-1017-RTX-PRO-Preview 41.6
  9. 9 deepseek-ai-DeepSeek-V4-Flash-0731-GB10-Preview 41.0
  10. 10 Grok 4.5 39.6
  11. 11 Gemini 3.5 Flash 39.2
  12. 12 Muse Spark 1.1 39.1
  13. 13 Gemma 4 31B IT 39.0
  14. 14 unsloth-Kimi-K3-GGUF-UD-IQ2_XXS-Reasoning-High 36.9
  15. 15 Inkling 36.3
  16. 16 DeepSeek-V4-Flash 35.9
  17. 17 deepseek-ai-DeepSeek-V4-Flash-RTX-PRO 35.4
  18. 18 unsloth-GLM-5.2-UD-IQ4_XS 35.3
  19. 19 Qwen3.7-Plus 35.0
  20. 20 qwen3.6-27b-mtp@q6_k 31.2
  21. 21 deepseek-ai-DeepSeek-V4-Flash 29.9
  22. 22 Kimi K2.7 Code 29.7
  23. 23 GLM-5.3-Flash-NVFP4-GB10-Preview 29.6
  24. 24 nvidia/nemotron-3-ultra-550b-a55b 26.4
  25. 25 GLM-5.3-Flash-NVFP4-RTX-PRO-Preview 25.2
  26. 26 GLM-5.2 23.8
  27. 27 NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 23.3
  28. 28 unsloth-Qwen3.6-27B-MTP-GGUF-Q8_0 21.7
  29. 29 Nex-N2-Pro 19.6
  30. 30 qwen3.6-27b-int8-havenoammo 19.3
  31. 31 Qwen3.8-27B-Q4_K_M 18.8
  32. 32 Step 3.7 Flash 18.7
  33. 33 google/gemma-4-26b-a4b-qat 18.3
  34. 34 qwen3.6-27b-autoround 18.1
  35. 35 Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M 17.5
  36. 36 Qwen-Qwen3.8-27B-FP8-RTX-PRO-Preview 17.2
  37. 37 MiniMax-M3 15.0
  38. 38 Qwen-AgentWorld-35B-A3B-UD-Q4_K_M 13.3
  39. 39 gpt-oss-120b 13.3
  40. 40 unsloth-Qwen3.6-35B-A3B-MTP-GGUF-Q8_K_XL 11.3
  41. 41 GLM-4.6V-Flash 11.0
  42. 42 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16 10.3
  43. 43 unsloth-GLM-5.2-UD-Q6-K-XL 9.80
  44. 44 Qwen 3.6 35B-A3B 8.50
  45. 45 Tencent Hy3 8.50
  46. 46 NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Q4_K_M 7.50
  47. 47 nvidia/GLM-5.1-NVFP4 7.00
  48. 48 Qwen 3.8 27B 6.50
  49. 49 cyankiwi/MiniMax-M2.7-AWQ-4bit 5.50
  50. 50 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16.gguf 5.50
  51. 51 gpt-oss-20b 4.80
  52. 52 MiniMax-M2.7 4.80
  53. 53 unsloth-GLM-4.6V-UD-Q4_K_XL 4.70
  54. 54 LFM2.5-8B-A1B 4.20
  55. 55 /data/models/safetensors/MiniMaxAI/MiniMax-M2.7 3.30
  56. 56 Ministral 3 3B Instruct 2.50
  57. 57 GPT-5.6 Terra 2.40
  58. 58 Gemma 4 26B A4B 2.20
  59. 59 openai/gpt-5.6-terra-pro 2.10
  60. 60 lfm2.5-350m 0.70
  61. 61 Qwen3.5-4B 0.30
  62. 62 GPT-5.6 Sol 0.10

Source: Official Evalatro v2 leaderboard

#EntryScoreSettingDateSource
1 GLM-5.3-Flash-NVFP4-Spark-RTX-PRO-Preview 46.2 Official leaderboard
2 DeepSeek-V4.1-Flash-1017-RTX-PRO-Max-Thinking 46.1 Official leaderboard
3 MiMo-V2.5-Pro 46.1 Apr 2026 Official leaderboard
4 Qwen3.7-Max 45.5 May 2026 Official leaderboard
5 deepseek-ai-DeepSeek-V4-Flash-0731-RTX-PRO-Preview 43.5 Official leaderboard
6 DeepSeek-V4-Pro 43.2 Apr 2026 Official leaderboard
7 Gemini 3.1 Pro 41.8 Feb 2026 Official leaderboard
8 DeepSeek-V4.1-Flash-1017-RTX-PRO-Preview 41.6 Official leaderboard
9 deepseek-ai-DeepSeek-V4-Flash-0731-GB10-Preview 41.0 Official leaderboard
10 Grok 4.5 39.6 Jul 2026 Official leaderboard
11 Gemini 3.5 Flash 39.2 May 2026 Official leaderboard
12 Muse Spark 1.1 39.1 Jul 2026 Official leaderboard
13 Gemma 4 31B IT 39.0 Apr 2026 Official leaderboard
14 unsloth-Kimi-K3-GGUF-UD-IQ2_XXS-Reasoning-High 36.9 Official leaderboard
15 Inkling 36.3 Jul 2026 Official leaderboard
16 DeepSeek-V4-Flash 35.9 Apr 2026 Official leaderboard
17 deepseek-ai-DeepSeek-V4-Flash-RTX-PRO 35.4 Official leaderboard
18 unsloth-GLM-5.2-UD-IQ4_XS 35.3 Official leaderboard
19 Qwen3.7-Plus 35.0 Jun 2026 Official leaderboard
20 qwen3.6-27b-mtp@q6_k 31.2 Official leaderboard
21 deepseek-ai-DeepSeek-V4-Flash 29.9 Official leaderboard
22 Kimi K2.7 Code 29.7 Jun 2026 Official leaderboard
23 GLM-5.3-Flash-NVFP4-GB10-Preview 29.6 Official leaderboard
24 nvidia/nemotron-3-ultra-550b-a55b 26.4 Official leaderboard
25 GLM-5.3-Flash-NVFP4-RTX-PRO-Preview 25.2 Official leaderboard
26 GLM-5.2 23.8 Jun 2026 Official leaderboard
27 NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 23.3 Official leaderboard
28 unsloth-Qwen3.6-27B-MTP-GGUF-Q8_0 21.7 Official leaderboard
29 Nex-N2-Pro 19.6 Official leaderboard
30 qwen3.6-27b-int8-havenoammo 19.3 Official leaderboard
31 Qwen3.8-27B-Q4_K_M 18.8 Official leaderboard
32 Step 3.7 Flash 18.7 May 2026 Official leaderboard
33 google/gemma-4-26b-a4b-qat 18.3 Official leaderboard
34 qwen3.6-27b-autoround 18.1 Official leaderboard
35 Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M 17.5 Official leaderboard
36 Qwen-Qwen3.8-27B-FP8-RTX-PRO-Preview 17.2 Official leaderboard
37 MiniMax-M3 15.0 Jun 2026 Official leaderboard
38 Qwen-AgentWorld-35B-A3B-UD-Q4_K_M 13.3 Official leaderboard
39 gpt-oss-120b 13.3 Aug 2025 Official leaderboard
40 unsloth-Qwen3.6-35B-A3B-MTP-GGUF-Q8_K_XL 11.3 Official leaderboard
41 GLM-4.6V-Flash 11.0 Dec 2025 Official leaderboard
42 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16 10.3 Official leaderboard
43 unsloth-GLM-5.2-UD-Q6-K-XL 9.80 Official leaderboard
44 Qwen 3.6 35B-A3B 8.50 Apr 2026 Official leaderboard
45 Tencent Hy3 8.50 Jul 2026 Official leaderboard
46 NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Q4_K_M 7.50 Official leaderboard
47 nvidia/GLM-5.1-NVFP4 7.00 Official leaderboard
48 Qwen 3.8 27B 6.50 Aug 2026 Official leaderboard
49 cyankiwi/MiniMax-M2.7-AWQ-4bit 5.50 Official leaderboard
50 Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-BF16.gguf 5.50 Official leaderboard
51 gpt-oss-20b 4.80 Aug 2025 Official leaderboard
52 MiniMax-M2.7 4.80 Mar 2026 Official leaderboard
53 unsloth-GLM-4.6V-UD-Q4_K_XL 4.70 Official leaderboard
54 LFM2.5-8B-A1B 4.20 May 2026 Official leaderboard
55 /data/models/safetensors/MiniMaxAI/MiniMax-M2.7 3.30 Official leaderboard
56 Ministral 3 3B Instruct 2.50 Dec 2025 Official leaderboard
57 GPT-5.6 Terra 2.40 Jul 2026 Official leaderboard
58 Gemma 4 26B A4B 2.20 Apr 2026 Official leaderboard
59 openai/gpt-5.6-terra-pro 2.10 Official leaderboard
60 lfm2.5-350m 0.70 Official leaderboard
61 Qwen3.5-4B 0.30 Feb 2026 Official leaderboard
62 GPT-5.6 Sol 0.10 Jul 2026 Official leaderboard
#ModelScoreSettingDateSource
1 MiMo-V2.5-Pro 46.1 Apr 2026 Official leaderboard
2 Qwen3.7-Max 45.5 May 2026 Official leaderboard
3 DeepSeek-V4-Pro 43.2 Apr 2026 Official leaderboard
4 Gemini 3.1 Pro 41.8 Feb 2026 Official leaderboard
5 Grok 4.5 39.6 Jul 2026 Official leaderboard
6 Gemini 3.5 Flash 39.2 May 2026 Official leaderboard
7 Muse Spark 1.1 39.1 Jul 2026 Official leaderboard
8 Gemma 4 31B IT 39.0 Apr 2026 Official leaderboard
9 Inkling 36.3 Jul 2026 Official leaderboard
10 DeepSeek-V4-Flash 35.9 Apr 2026 Official leaderboard
11 Qwen3.7-Plus 35.0 Jun 2026 Official leaderboard
12 Kimi K2.7 Code 29.7 Jun 2026 Official leaderboard
13 GLM-5.2 23.8 Jun 2026 Official leaderboard
14 Step 3.7 Flash 18.7 May 2026 Official leaderboard
15 MiniMax-M3 15.0 Jun 2026 Official leaderboard
16 gpt-oss-120b 13.3 Aug 2025 Official leaderboard
17 GLM-4.6V-Flash 11.0 Dec 2025 Official leaderboard
18 Qwen 3.6 35B-A3B 8.50 Apr 2026 Official leaderboard
19 Tencent Hy3 8.50 Jul 2026 Official leaderboard
20 Qwen 3.8 27B 6.50 Aug 2026 Official leaderboard
21 gpt-oss-20b 4.80 Aug 2025 Official leaderboard
22 MiniMax-M2.7 4.80 Mar 2026 Official leaderboard
23 LFM2.5-8B-A1B 4.20 May 2026 Official leaderboard
24 Ministral 3 3B Instruct 2.50 Dec 2025 Official leaderboard
25 GPT-5.6 Terra 2.40 Jul 2026 Official leaderboard
26 Gemma 4 26B A4B 2.20 Apr 2026 Official leaderboard
27 Qwen3.5-4B 0.30 Feb 2026 Official leaderboard
28 GPT-5.6 Sol 0.10 Jul 2026 Official leaderboard

Number one over time

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

Xiaomi Corp
Own leaderboard Not tested since Aug 2026

Source: Official Evalatro v2 leaderboard

Scores over time

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

Plot with other benchmarks
On the benchmark’s own leaderboard
0.00 10.0 20.0 30.0 40.0 50.0 2026 0.00 10.0 20.0 30.0 40.0 50.0 2026

Source: Official Evalatro v2 leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Games and long tasks

All 261 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (62 entries).
  • Pass Models scored 28 models we track, plus 34 other entries on its leaderboard.
  • Pass Newest score The newest is from August 2026, for Qwen 3.8 27B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.6 Sol (#14).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on evalatro.dev: 29 rows for models we track and 34 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check