Benchmark Almanac

EuroEval Faroese Grammatical Error Correction

Gemma 3 27B PT 16.9% leads since August 2026
Not in model cards

Models rewrite a Faroese student’s sentence to correct its grammar. An exact match to the reference correction earns credit.

EuroEval lists test scores for 245 distinct model names on 2,048 sentences. The original paper asks models to choose between two sentences.

Released
August 2026
Kind
Reading in many languages
Level
School
Scored
Checked automatically
Answers
Short answer
Tasks
Kept private
Measures
Capability
Language
Faroese
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
A host
Strongest model tested
QwQ-32B #137 on the capability index
Last model added
January 2026
Scores refreshed
November 20th, 2026
241
Entries on its own leaderboard, 71 models we track

Gemma 3 27B PT leads with 16.9% among models released by January 2026 Gemma 3 27B PT leads with 16.9% among models released by January 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. This view shows only models we track, each at its best.

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

241 entries. Scroll sideways for the rest.

  1. 16.9 Gemma 3 27B PT
  2. 16.1 norallm/normistral-11b-warm
  3. 16.0 norallm/normistral-11b-long
  4. 14.2 speakleash/Bielik-11B-v3-Base-20250730
  5. 13.8 LumiOpen/Viking-13B
  6. 13.7 Apertus 70B
  7. 13.0 LumiOpen/Viking-33B
  8. 12.9 Gemma 3 12B PT
  9. 11.8 Cogito v1 Preview 70B
  10. 11.2 swiss-ai/Apertus-8B-2509
  11. 11.1 TrustLLMeu/baseline-7-8b_2-3t-tokens_llama
  12. 11.0 TrustLLMeu/trustllm-v2-8b-midtrain@3c7af30bc43bb3c8607e0fab6e7f2a23dd64af49
  13. 10.7 deepcogito/cogito-v2-preview-llama-70B
  14. 10.2 TrustLLMeu/trustllm-v2-8b@d8cc924fa550fbf9474f5251e8e2f92d1c1bf894
  15. 10.1 Llama 3.1-70B
  16. 9.84 Llama 4 Scout
  17. 9.57 LumiOpen/Llama-Poro-2-70B-SFT
  18. 8.53 TrustLLMeu/baseline-7-8b_1t-tokens_llama
  19. 8.33 Apertus 8B Instruct
  20. 8.23 baidu/ERNIE-4.5-21B-A3B-PT
  21. 7.11 Gemma 3 12B
  22. 7.01 Mistral Small 3.2
  23. 7.00 Mistral Small 3.1
  24. 6.61 Gemma 7B
  25. 6.29 Llama 3.1-8B
  26. 6.27 mistralai/Mistral-Small-3.1-24B-Base-2503
  27. 6.10 Gemma 2 9B
  28. 6.10 Gemma 3 4B PT
  29. 5.92 deepcogito/cogito-v1-preview-llama-8B
  30. 5.59 TrustLLMeu/trustllm-v2-8b-sft-8k@2aadff8653393e41abf583222164e279461a619f
  31. 5.50 TrustLLMeu/trustllm-v2-8b-sft@f442f651172c1a97143738fc30c3a6342d1e8660
  32. 5.48 TrustLLMeu/trustllm-v2-8b-sft-mixb2-exp@f33c260efa853437a55d780ce5b910944b1395f6
  33. 5.14 norallm/normistral-11b-thinking
  34. 5.12 allenai/Llama-3.1-Tulu-3-8B-SFT
  35. 5.06 mistralai/Mistral-Small-24B-Instruct-2501
  36. 5.00 meta-llama/Llama-2-70b-hf
  37. 5.00 mistralai/Magistral-Small-2509
  38. 4.80 deepcogito/cogito-v1-preview-qwen-32B
  39. 4.72 Qwen/Qwen3-14B#no-thinking
  40. 4.51 Qwen/Qwen3-235B-A22B-Instruct-2507-FP8
  41. 4.44 allenai/Llama-3.1-Tulu-3.1-8B
  42. 4.44 Qwen3-Next-80B-A3B-Instruct
  43. 4.31 google/gemma-2-9b-it
  44. 4.30 LumiOpen/Llama-Poro-2-70B-Instruct
  45. 4.07 CohereForAI/aya-23-35B
  46. 4.06 NousResearch/DeepHermes-3-Llama-3-8B-Preview
  47. 4.05 allenai/Llama-3.1-Tulu-3-8B-DPO
  48. 4.02 Llama-3.1-Tulu-3-8B
  49. 3.49 Gemma 3 4B
  50. 3.29 deepcogito/cogito-v1-preview-qwen-14B
  51. 3.29 Yi-1.5-34B
  52. 3.23 gpt-oss-20b
  53. 3.17 allenai/OLMo-2-1124-13B
  54. 3.08 meta-llama/Llama-2-13b-hf
  55. 2.95 allenai/Olmo-3-1125-32B
  56. 2.93 utter-project/EuroLLM-22B-2512
  57. 2.92 meta-llama/Llama-3.1-8B-Instruct
  58. 2.87 mistralai/Mistral-Nemo-Instruct-2407
  59. 2.82 ai21labs/Jamba-v0.1
  60. 2.80 deepcogito/cogito-v1-preview-llama-3B
  61. 2.74 utter-project/EuroLLM-22B-Instruct-2512
  62. 2.63 allenai/OLMo-2-0325-32B
  63. 2.63 Qwen/Qwen3-32B#no-thinking
  64. 2.62 upstage/SOLAR-10.7B-v1.0
  65. 2.56 Phi-4
  66. 2.54 01-ai/Yi-1.5-9B
  67. 2.32 Qwen/Qwen3-30B-A3B#no-thinking
  68. 2.30 allenai/Olmo-3.1-32B-Instruct-SFT
  69. 2.30 NbAiLab/borealis-4b-instruct-preview
  70. 2.28 utter-project/EuroLLM-22B-Preview
  71. 2.15 speakleash/Bielik-4.5B-v3.0-Instruct
  72. 2.11 mistralai/Ministral-3-3B-Base-2512
  73. 2.08 ibm-granite/granite-20b-code-base
  74. 2.06 mistralai/Ministral-3-8B-Reasoning-2512
  75. 1.97 DeepSeek-R1-Distill-Llama-70B
  76. 1.93 ibm-granite/granite-8b-code-base-4k
  77. 1.92 NovaSky-AI/Sky-T1-32B-Preview
  78. 1.91 meta-llama/Llama-2-13b-chat-hf
  79. 1.90 HuggingFaceTB/SmolLM3-3B#no-thinking
  80. 1.90 speakleash/Bielik-11B-v2.6-Instruct
  81. 1.83 Granite-3.0-8B-Base
  82. 1.82 utter-project/EuroLLM-9B-Instruct
  83. 1.80 Granite-3.0-8B-Instruct
  84. 1.78 Qwen/Qwen3-8B#no-thinking
  85. 1.72 arcee-ai/Trinity-Mini-Base
  86. 1.70 Gemma 2 2B IT
  87. 1.68 ibm-granite/granite-34b-code-instruct
  88. 1.64 utter-project/EuroLLM-9B
  89. 1.63 Qwen/Qwen3-1.7B#no-thinking
  90. 1.61 meta-llama/Llama-3.2-3B-Instruct
  91. 1.59 ibm-granite/granite-34b-code-base
  92. 1.57 allenai/Olmo-3-1025-7B
  93. 1.57 speakleash/Bielik-11B-v2.3-Instruct
  94. 1.53 Command R7B
  95. 1.51 meta-llama/Llama-2-7b-chat-hf
  96. 1.51 Aya Expanse 32B
  97. 1.50 infly/OpenCoder-8B-Instruct
  98. 1.49 neuralmagic/Sparse-Llama-3.1-8B-2of4
  99. 1.47 tiiuae/Falcon3-10B-Base
  100. 1.43 occiglot/occiglot-7b-eu5-instruct
  101. 1.37 allenai/OLMo-2-1124-7B-Instruct
  102. 1.36 Gemma 2B
  103. 1.35 ibm-granite/granite-8b-code-instruct-4k
  104. 1.31 Falcon3-10B-Instruct
  105. 1.30 ibm-granite/granite-20b-code-instruct
  106. 1.24 Qwen/Qwen3-4B#no-thinking
  107. 1.23 baidu/ERNIE-4.5-0.3B-PT
  108. 1.22 Qwen1.5-14B
  109. 1.20 Qwen/Qwen3-0.6B#no-thinking
  110. 1.19 utter-project/EuroLLM-22B-Instruct-Preview
  111. 1.14 allenai/Olmo-3-7B-Instruct-SFT
  112. 1.03 allenai/OLMo-2-1124-7B
  113. 1.01 Qwen/Qwen2.5-Math-7B
  114. 1.00 google/gemma-2-2b
  115. 0.99 LiquidAI/LFM2-2.6B
  116. 0.97 Ministral 3 3B Reasoning
  117. 0.96 Reka Flash 3
  118. 0.95 Llama 3.2 1B
  119. 0.89 arcee-ai/AFM-4.5B-Base
  120. 0.89 Falcon3-7B-Instruct
  121. 0.88 Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
  122. 0.87 qwen3-4b-instruct-2507
  123. 0.84 tiiuae/Falcon3-7B-Base
  124. 0.84 Qwen3-14B
  125. 0.82 common-pile/comma-v0.1-2t
  126. 0.80 speakleash/Bielik-1.5B-v3.0-Instruct
  127. 0.79 AFM-4.5B
  128. 0.78 PleIAs/Pleias-1.2b-Preview
  129. 0.78 Reka Flash 3.1
  130. 0.77 LFM2-1.2B
  131. 0.75 bineric/lynx-instruct-30b#no-thinking
  132. 0.74 bineric/lynx-instruct-30b#thinking
  133. 0.73 facebook/MobileLLM-R1-950M-base
  134. 0.73 infly/OpenCoder-8B-Base
  135. 0.68 ibm-granite/granite-3b-code-instruct-2k
  136. 0.67 SmolLM2 1.7B Instruct
  137. 0.65 ibm-granite/granite-3.0-3b-a800m-base
  138. 0.63 meta-llama/Llama-3.2-1B-Instruct
  139. 0.62 LiquidAI/LFM2-700M
  140. 0.62 mii-llm/nesso-4B
  141. 0.62 nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16#no-thinking
  142. 0.60 infly/OpenCoder-1.5B-Instruct
  143. 0.57 occiglot/occiglot-7b-eu5
  144. 0.56 OLMo 2 32B Instruct
  145. 0.55 HuggingFaceTB/SmolLM3-3B-Base
  146. 0.54 ibm-granite/granite-3b-code-base-2k
  147. 0.54 Qwen/Qwen2.5-1.5B-Instruct
  148. 0.53 Falcon3-3B-Instruct
  149. 0.52 Gemma 3 1B
  150. 0.52 Qwen3-Next-80B-A3B
  151. 0.51 HuggingFaceTB/SmolLM2-135M-Instruct
  152. 0.51 Gemma 3 1B PT
  153. 0.48 allenai/OLMo-2-1124-13B-Instruct
  154. 0.46 Olmo 3.1 32B Instruct
  155. 0.45 ibm-granite/granite-3.0-1b-a400m-instruct
  156. 0.45 openGPT-X/Teuken-7B-instruct-commercial-v0.4
  157. 0.43 152334H/miqu-1-70b-sf
  158. 0.43 common-pile/comma-v0.1-1t
  159. 0.42 LiquidAI/LFM2-350M
  160. 0.41 utter-project/EuroVLM-1.7B-Preview
  161. 0.40 Qwen3-1.7B
  162. 0.38 senseable/WestLake-7B-v2
  163. 0.36 SmolLM2-1.7B
  164. 0.35 LiquidAI/LFM2.5-1.2B-Base
  165. 0.34 allenai/Olmo-3-7B-Instruct
  166. 0.34 allenai/Olmo-3.1-32B-Instruct-DPO
  167. 0.34 Qwen/Qwen1.5-0.5B-Chat
  168. 0.33 kyutai/helium-1-2b
  169. 0.33 meta-llama/Llama-4-Scout-17B-16E
  170. 0.32 utter-project/EuroMoE-2.6B-A0.6B-Instruct-Preview
  171. 0.32 Phi-2
  172. 0.32 QwQ-32B
  173. 0.32 Qwen3-4B
  174. 0.30 google/gemma-2b-it
  175. 0.29 Gemma 3 270M
  176. 0.26 HuggingFaceTB/SmolLM2-360M-Instruct
  177. 0.25 facebook/MobileLLM-R1-360M-base
  178. 0.23 meta-llama/Llama-2-70b-chat-hf
  179. 0.22 allenai/Olmo-3-7B-Instruct-DPO
  180. 0.22 ibm-granite/granite-3.0-1b-a400m-base
  181. 0.22 DeepSeek-R1-Distill-Qwen-32B
  182. 0.21 tiiuae/Falcon3-3B-Base
  183. 0.21 SmolLM2-135M
  184. 0.21 Phi-4-Reasoning-plus
  185. 0.21 LFM2.5-1.2B-Instruct
  186. 0.20 BSC-LT/salamandra-7b
  187. 0.20 deepseek-ai/DeepSeek-R1-Distill-Llama-8B
  188. 0.20 infly/OpenCoder-1.5B-Base
  189. 0.19 Phi-4-reasoning
  190. 0.18 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Instruct
  191. 0.16 DeepSeek-R1-Distill-Qwen-14B
  192. 0.15 HPLT/hplt2c_isl_checkpoints
  193. 0.13 Qwen1.5-14B-Chat
  194. 0.13 nvidia-nemotron-nano-9b-v2
  195. 0.12 facebook/MobileLLM-R1-140M-base
  196. 0.11 Qwen/Qwen1.5-0.5B
  197. 0.10 HPLT/hplt2c_nob_checkpoints
  198. 0.10 openai-community/gpt2-large
  199. 0.09 BSC-LT/salamandra-7b-instruct
  200. 0.09 SmolLM2-360M
  201. 0.09 Qwen3-0.6B
  202. 0.08 google/gemma-7b-it
  203. 0.08 HPLT/hplt2c_dan_checkpoints
  204. 0.08 ThatsGroes/gemma-2-27b-it-FP8-Dynamic
  205. 0.07 google/gemma-2-27b-it
  206. 0.06 HPLT/hplt2c_swe_checkpoints
  207. 0.06 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Base
  208. 0.05 HPLT/hplt2c_nno_checkpoints
  209. 0.05 PleIAs/Pleias-350m-Preview
  210. 0.04 nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16
  211. 0.03 Qwen3-4B-Thinking-2507
  212. 0.02 yulan-team/YuLan-Mini-Instruct
  213. 0.01 openai-community/gpt2-medium
  214. 0.0 allenai/Olmo-3-7B-Think
  215. 0.0 BSC-LT/ALIA-40b
  216. 0.0 DiscoResearch/DiscoLM-70b
  217. 0.0 facebook/MobileLLM-R1-140M
  218. 0.0 facebook/MobileLLM-R1-360M
  219. 0.0 facebook/MobileLLM-R1-950M
  220. 0.0 HuggingFaceTB/SmolLM3-3B
  221. 0.0 MaLA-LM/emma-500-llama2-7b
  222. 0.0 nvidia/NVIDIA-Nemotron-Nano-9B-v2#no-thinking
  223. 0.0 openai-community/gpt2
  224. 0.0 PleIAs/Baguettotron@refs/pr/6#no-thinking
  225. 0.0 PleIAs/Monad
  226. 0.0 Qwen/Qwen2.5-Math-1.5B
  227. 0.0 Shekswess/trlm-135m
  228. 0.0 tencent/Hunyuan-1.8B-Instruct
  229. 0.0 tencent/Hunyuan-7B-Instruct
  230. 0.0 tiiuae/falcon-11B
  231. 0.0 TildeAI/TildeOpen-30b#slow-tokenizer
  232. 0.0 timpal0l/Mistral-7B-v0.1-flashback-v2
  233. 0.0 Mistral 7B v0.3
  234. 0.0 Gemma 2 27B
  235. 0.0 DeepSeek-R1-Distill-Qwen-1.5B
  236. 0.0 DeepSeek-R1-Distill-Qwen-7B
  237. 0.0 Qwen3-8B
  238. 0.0 Phi-4-mini-reasoning
  239. 0.0 deepseek-r1-0528-qwen3-8b
  240. 0.0 GLM-4.1V-9B-Thinking
  241. 0.0 LFM2.5-1.2B-Thinking
  1. 1 Gemma 3 27B PT 16.9%
  2. 2 norallm/normistral-11b-warm 16.1%
  3. 3 norallm/normistral-11b-long 16%
  4. 4 speakleash/Bielik-11B-v3-Base-20250730 14.2%
  5. 5 LumiOpen/Viking-13B 13.8%
  6. 6 Apertus 70B 13.7%
  7. 7 LumiOpen/Viking-33B 13%
  8. 8 Gemma 3 12B PT 12.9%
  9. 9 Cogito v1 Preview 70B 11.8%
  10. 10 swiss-ai/Apertus-8B-2509 11.2%
  11. 11 TrustLLMeu/baseline-7-8b_2-3t-tokens_llama 11.1%
  12. 12 TrustLLMeu/trustllm-v2-8b-midtrain@3c7af30bc43bb3c8607e0fab6e7f2a23dd64af49 11%
  13. 13 deepcogito/cogito-v2-preview-llama-70B 10.7%
  14. 14 TrustLLMeu/trustllm-v2-8b@d8cc924fa550fbf9474f5251e8e2f92d1c1bf894 10.2%
  15. 15 Llama 3.1-70B 10.1%
  16. 16 Llama 4 Scout 9.84%
  17. 17 LumiOpen/Llama-Poro-2-70B-SFT 9.57%
  18. 18 TrustLLMeu/baseline-7-8b_1t-tokens_llama 8.53%
  19. 19 Apertus 8B Instruct 8.33%
  20. 20 baidu/ERNIE-4.5-21B-A3B-PT 8.23%
  21. 21 Gemma 3 12B 7.11%
  22. 22 Mistral Small 3.2 7.01%
  23. 23 Mistral Small 3.1 7.00%
  24. 24 Gemma 7B 6.61%
  25. 25 Llama 3.1-8B 6.29%
  26. 26 mistralai/Mistral-Small-3.1-24B-Base-2503 6.27%
  27. 27 Gemma 2 9B 6.10%
  28. 28 Gemma 3 4B PT 6.10%
  29. 29 deepcogito/cogito-v1-preview-llama-8B 5.92%
  30. 30 TrustLLMeu/trustllm-v2-8b-sft-8k@2aadff8653393e41abf583222164e279461a619f 5.59%
  31. 31 TrustLLMeu/trustllm-v2-8b-sft@f442f651172c1a97143738fc30c3a6342d1e8660 5.50%
  32. 32 TrustLLMeu/trustllm-v2-8b-sft-mixb2-exp@f33c260efa853437a55d780ce5b910944b1395f6 5.48%
  33. 33 norallm/normistral-11b-thinking 5.14%
  34. 34 allenai/Llama-3.1-Tulu-3-8B-SFT 5.12%
  35. 35 mistralai/Mistral-Small-24B-Instruct-2501 5.06%
  36. 36 meta-llama/Llama-2-70b-hf 5.00%
  37. 37 mistralai/Magistral-Small-2509 5.00%
  38. 38 deepcogito/cogito-v1-preview-qwen-32B 4.80%
  39. 39 Qwen/Qwen3-14B#no-thinking 4.72%
  40. 40 Qwen/Qwen3-235B-A22B-Instruct-2507-FP8 4.51%
  41. 41 allenai/Llama-3.1-Tulu-3.1-8B 4.44%
  42. 42 Qwen3-Next-80B-A3B-Instruct 4.44%
  43. 43 google/gemma-2-9b-it 4.31%
  44. 44 LumiOpen/Llama-Poro-2-70B-Instruct 4.30%
  45. 45 CohereForAI/aya-23-35B 4.07%
  46. 46 NousResearch/DeepHermes-3-Llama-3-8B-Preview 4.06%
  47. 47 allenai/Llama-3.1-Tulu-3-8B-DPO 4.05%
  48. 48 Llama-3.1-Tulu-3-8B 4.02%
  49. 49 Gemma 3 4B 3.49%
  50. 50 deepcogito/cogito-v1-preview-qwen-14B 3.29%
  51. 51 Yi-1.5-34B 3.29%
  52. 52 gpt-oss-20b low 3.23%
  53. 53 allenai/OLMo-2-1124-13B 3.17%
  54. 54 meta-llama/Llama-2-13b-hf 3.08%
  55. 55 allenai/Olmo-3-1125-32B 2.95%
  56. 56 utter-project/EuroLLM-22B-2512 2.93%
  57. 57 meta-llama/Llama-3.1-8B-Instruct 2.92%
  58. 58 mistralai/Mistral-Nemo-Instruct-2407 2.87%
  59. 59 ai21labs/Jamba-v0.1 2.82%
  60. 60 deepcogito/cogito-v1-preview-llama-3B 2.80%
  61. 61 utter-project/EuroLLM-22B-Instruct-2512 2.74%
  62. 62 allenai/OLMo-2-0325-32B 2.63%
  63. 63 Qwen/Qwen3-32B#no-thinking 2.63%
  64. 64 upstage/SOLAR-10.7B-v1.0 2.62%
  65. 65 Phi-4 2.56%
  66. 66 01-ai/Yi-1.5-9B 2.54%
  67. 67 Qwen/Qwen3-30B-A3B#no-thinking 2.32%
  68. 68 allenai/Olmo-3.1-32B-Instruct-SFT 2.30%
  69. 69 NbAiLab/borealis-4b-instruct-preview 2.30%
  70. 70 utter-project/EuroLLM-22B-Preview 2.28%
  71. 71 speakleash/Bielik-4.5B-v3.0-Instruct 2.15%
  72. 72 mistralai/Ministral-3-3B-Base-2512 2.11%
  73. 73 ibm-granite/granite-20b-code-base 2.08%
  74. 74 mistralai/Ministral-3-8B-Reasoning-2512 2.06%
  75. 75 DeepSeek-R1-Distill-Llama-70B 1.97%
  76. 76 ibm-granite/granite-8b-code-base-4k 1.93%
  77. 77 NovaSky-AI/Sky-T1-32B-Preview 1.92%
  78. 78 meta-llama/Llama-2-13b-chat-hf 1.91%
  79. 79 HuggingFaceTB/SmolLM3-3B#no-thinking 1.90%
  80. 80 speakleash/Bielik-11B-v2.6-Instruct 1.90%
  81. 81 Granite-3.0-8B-Base 1.83%
  82. 82 utter-project/EuroLLM-9B-Instruct 1.82%
  83. 83 Granite-3.0-8B-Instruct 1.80%
  84. 84 Qwen/Qwen3-8B#no-thinking 1.78%
  85. 85 arcee-ai/Trinity-Mini-Base 1.72%
  86. 86 Gemma 2 2B IT 1.70%
  87. 87 ibm-granite/granite-34b-code-instruct 1.68%
  88. 88 utter-project/EuroLLM-9B 1.64%
  89. 89 Qwen/Qwen3-1.7B#no-thinking 1.63%
  90. 90 meta-llama/Llama-3.2-3B-Instruct 1.61%
  91. 91 ibm-granite/granite-34b-code-base 1.59%
  92. 92 allenai/Olmo-3-1025-7B 1.57%
  93. 93 speakleash/Bielik-11B-v2.3-Instruct 1.57%
  94. 94 Command R7B 1.53%
  95. 95 meta-llama/Llama-2-7b-chat-hf 1.51%
  96. 96 Aya Expanse 32B 1.51%
  97. 97 infly/OpenCoder-8B-Instruct 1.50%
  98. 98 neuralmagic/Sparse-Llama-3.1-8B-2of4 1.49%
  99. 99 tiiuae/Falcon3-10B-Base 1.47%
  100. 100 occiglot/occiglot-7b-eu5-instruct 1.43%
  101. 101 allenai/OLMo-2-1124-7B-Instruct 1.37%
  102. 102 Gemma 2B 1.36%
  103. 103 ibm-granite/granite-8b-code-instruct-4k 1.35%
  104. 104 Falcon3-10B-Instruct 1.31%
  105. 105 ibm-granite/granite-20b-code-instruct 1.30%
  106. 106 Qwen/Qwen3-4B#no-thinking 1.24%
  107. 107 baidu/ERNIE-4.5-0.3B-PT 1.23%
  108. 108 Qwen1.5-14B 1.22%
  109. 109 Qwen/Qwen3-0.6B#no-thinking 1.20%
  110. 110 utter-project/EuroLLM-22B-Instruct-Preview 1.19%
  111. 111 allenai/Olmo-3-7B-Instruct-SFT 1.14%
  112. 112 allenai/OLMo-2-1124-7B 1.03%
  113. 113 Qwen/Qwen2.5-Math-7B 1.01%
  114. 114 google/gemma-2-2b 1.00%
  115. 115 LiquidAI/LFM2-2.6B 0.99%
  116. 116 Ministral 3 3B Reasoning 0.97%
  117. 117 Reka Flash 3 0.96%
  118. 118 Llama 3.2 1B 0.95%
  119. 119 arcee-ai/AFM-4.5B-Base 0.89%
  120. 120 Falcon3-7B-Instruct 0.89%
  121. 121 Qwen/Qwen3-235B-A22B-Thinking-2507-FP8 0.88%
  122. 122 qwen3-4b-instruct-2507 0.87%
  123. 123 tiiuae/Falcon3-7B-Base 0.84%
  124. 124 Qwen3-14B 0.84%
  125. 125 common-pile/comma-v0.1-2t 0.82%
  126. 126 speakleash/Bielik-1.5B-v3.0-Instruct 0.80%
  127. 127 AFM-4.5B 0.79%
  128. 128 PleIAs/Pleias-1.2b-Preview 0.78%
  129. 129 Reka Flash 3.1 0.78%
  130. 130 LFM2-1.2B 0.77%
  131. 131 bineric/lynx-instruct-30b#no-thinking 0.75%
  132. 132 bineric/lynx-instruct-30b#thinking 0.74%
  133. 133 facebook/MobileLLM-R1-950M-base 0.73%
  134. 134 infly/OpenCoder-8B-Base 0.73%
  135. 135 ibm-granite/granite-3b-code-instruct-2k 0.68%
  136. 136 SmolLM2 1.7B Instruct 0.67%
  137. 137 ibm-granite/granite-3.0-3b-a800m-base 0.65%
  138. 138 meta-llama/Llama-3.2-1B-Instruct 0.63%
  139. 139 LiquidAI/LFM2-700M 0.62%
  140. 140 mii-llm/nesso-4B 0.62%
  141. 141 nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16#no-thinking 0.62%
  142. 142 infly/OpenCoder-1.5B-Instruct 0.60%
  143. 143 occiglot/occiglot-7b-eu5 0.57%
  144. 144 OLMo 2 32B Instruct 0.56%
  145. 145 HuggingFaceTB/SmolLM3-3B-Base 0.55%
  146. 146 ibm-granite/granite-3b-code-base-2k 0.54%
  147. 147 Qwen/Qwen2.5-1.5B-Instruct 0.54%
  148. 148 Falcon3-3B-Instruct 0.53%
  149. 149 Gemma 3 1B 0.52%
  150. 150 Qwen3-Next-80B-A3B thinking 0.52%
  151. 151 HuggingFaceTB/SmolLM2-135M-Instruct 0.51%
  152. 152 Gemma 3 1B PT 0.51%
  153. 153 allenai/OLMo-2-1124-13B-Instruct 0.48%
  154. 154 Olmo 3.1 32B Instruct 0.46%
  155. 155 ibm-granite/granite-3.0-1b-a400m-instruct 0.45%
  156. 156 openGPT-X/Teuken-7B-instruct-commercial-v0.4 0.45%
  157. 157 152334H/miqu-1-70b-sf 0.43%
  158. 158 common-pile/comma-v0.1-1t 0.43%
  159. 159 LiquidAI/LFM2-350M 0.42%
  160. 160 utter-project/EuroVLM-1.7B-Preview 0.41%
  161. 161 Qwen3-1.7B 0.40%
  162. 162 senseable/WestLake-7B-v2 0.38%
  163. 163 SmolLM2-1.7B 0.36%
  164. 164 LiquidAI/LFM2.5-1.2B-Base 0.35%
  165. 165 allenai/Olmo-3-7B-Instruct 0.34%
  166. 166 allenai/Olmo-3.1-32B-Instruct-DPO 0.34%
  167. 167 Qwen/Qwen1.5-0.5B-Chat 0.34%
  168. 168 kyutai/helium-1-2b 0.33%
  169. 169 meta-llama/Llama-4-Scout-17B-16E 0.33%
  170. 170 utter-project/EuroMoE-2.6B-A0.6B-Instruct-Preview 0.32%
  171. 171 Phi-2 0.32%
  172. 172 QwQ-32B 0.32%
  173. 173 Qwen3-4B 0.32%
  174. 174 google/gemma-2b-it 0.30%
  175. 175 Gemma 3 270M 0.29%
  176. 176 HuggingFaceTB/SmolLM2-360M-Instruct 0.26%
  177. 177 facebook/MobileLLM-R1-360M-base 0.25%
  178. 178 meta-llama/Llama-2-70b-chat-hf 0.23%
  179. 179 allenai/Olmo-3-7B-Instruct-DPO 0.22%
  180. 180 ibm-granite/granite-3.0-1b-a400m-base 0.22%
  181. 181 DeepSeek-R1-Distill-Qwen-32B 0.22%
  182. 182 tiiuae/Falcon3-3B-Base 0.21%
  183. 183 SmolLM2-135M 0.21%
  184. 184 Phi-4-Reasoning-plus 0.21%
  185. 185 LFM2.5-1.2B-Instruct 0.21%
  186. 186 BSC-LT/salamandra-7b 0.20%
  187. 187 deepseek-ai/DeepSeek-R1-Distill-Llama-8B 0.20%
  188. 188 infly/OpenCoder-1.5B-Base 0.20%
  189. 189 Phi-4-reasoning 0.19%
  190. 190 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Instruct 0.18%
  191. 191 DeepSeek-R1-Distill-Qwen-14B 0.16%
  192. 192 HPLT/hplt2c_isl_checkpoints 0.15%
  193. 193 Qwen1.5-14B-Chat 0.13%
  194. 194 nvidia-nemotron-nano-9b-v2 0.13%
  195. 195 facebook/MobileLLM-R1-140M-base 0.12%
  196. 196 Qwen/Qwen1.5-0.5B 0.11%
  197. 197 HPLT/hplt2c_nob_checkpoints 0.10%
  198. 198 openai-community/gpt2-large 0.10%
  199. 199 BSC-LT/salamandra-7b-instruct 0.09%
  200. 200 SmolLM2-360M 0.09%
  201. 201 Qwen3-0.6B 0.09%
  202. 202 google/gemma-7b-it 0.08%
  203. 203 HPLT/hplt2c_dan_checkpoints 0.08%
  204. 204 ThatsGroes/gemma-2-27b-it-FP8-Dynamic 0.08%
  205. 205 google/gemma-2-27b-it 0.07%
  206. 206 HPLT/hplt2c_swe_checkpoints 0.06%
  207. 207 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Base 0.06%
  208. 208 HPLT/hplt2c_nno_checkpoints 0.05%
  209. 209 PleIAs/Pleias-350m-Preview 0.05%
  210. 210 nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16 0.04%
  211. 211 Qwen3-4B-Thinking-2507 0.03%
  212. 212 yulan-team/YuLan-Mini-Instruct 0.02%
  213. 213 openai-community/gpt2-medium 0.01%
  214. 214 allenai/Olmo-3-7B-Think 0%
  215. 215 BSC-LT/ALIA-40b 0%
  216. 216 DiscoResearch/DiscoLM-70b 0%
  217. 217 facebook/MobileLLM-R1-140M 0%
  218. 218 facebook/MobileLLM-R1-360M 0%
  219. 219 facebook/MobileLLM-R1-950M 0%
  220. 220 HuggingFaceTB/SmolLM3-3B 0%
  221. 221 MaLA-LM/emma-500-llama2-7b 0%
  222. 222 nvidia/NVIDIA-Nemotron-Nano-9B-v2#no-thinking 0%
  223. 223 openai-community/gpt2 0%
  224. 224 PleIAs/Baguettotron@refs/pr/6#no-thinking 0%
  225. 225 PleIAs/Monad 0%
  226. 226 Qwen/Qwen2.5-Math-1.5B 0%
  227. 227 Shekswess/trlm-135m 0%
  228. 228 tencent/Hunyuan-1.8B-Instruct 0%
  229. 229 tencent/Hunyuan-7B-Instruct 0%
  230. 230 tiiuae/falcon-11B 0%
  231. 231 TildeAI/TildeOpen-30b#slow-tokenizer 0%
  232. 232 timpal0l/Mistral-7B-v0.1-flashback-v2 0%
  233. 233 Mistral 7B v0.3 0%
  234. 234 Gemma 2 27B 0%
  235. 235 DeepSeek-R1-Distill-Qwen-1.5B 0%
  236. 236 DeepSeek-R1-Distill-Qwen-7B 0%
  237. 237 Qwen3-8B 0%
  238. 238 Phi-4-mini-reasoning 0%
  239. 239 deepseek-r1-0528-qwen3-8b 0%
  240. 240 GLM-4.1V-9B-Thinking 0%
  241. 241 LFM2.5-1.2B-Thinking 0%

Source: Official EuroEval Faroese Grammatical Error Correction leaderboard

71 models. Scroll sideways for the rest.

  1. 16.9 Gemma 3 27B PT
  2. 13.7 Apertus 70B
  3. 12.9 Gemma 3 12B PT
  4. 11.8 Cogito v1 Preview 70B
  5. 10.1 Llama 3.1-70B
  6. 9.84 Llama 4 Scout
  7. 8.33 Apertus 8B Instruct
  8. 7.11 Gemma 3 12B
  9. 7.01 Mistral Small 3.2
  10. 7.00 Mistral Small 3.1
  11. 6.61 Gemma 7B
  12. 6.29 Llama 3.1-8B
  13. 6.10 Gemma 2 9B
  14. 6.10 Gemma 3 4B PT
  15. 4.44 Qwen3-Next-80B-A3B-Instruct
  16. 4.02 Llama-3.1-Tulu-3-8B
  17. 3.49 Gemma 3 4B
  18. 3.29 Yi-1.5-34B
  19. 3.23 gpt-oss-20b
  20. 2.56 Phi-4
  21. 1.97 DeepSeek-R1-Distill-Llama-70B
  22. 1.83 Granite-3.0-8B-Base
  23. 1.80 Granite-3.0-8B-Instruct
  24. 1.70 Gemma 2 2B IT
  25. 1.53 Command R7B
  26. 1.51 Aya Expanse 32B
  27. 1.36 Gemma 2B
  28. 1.31 Falcon3-10B-Instruct
  29. 1.22 Qwen1.5-14B
  30. 0.97 Ministral 3 3B Reasoning
  31. 0.96 Reka Flash 3
  32. 0.95 Llama 3.2 1B
  33. 0.89 Falcon3-7B-Instruct
  34. 0.87 qwen3-4b-instruct-2507
  35. 0.84 Qwen3-14B
  36. 0.79 AFM-4.5B
  37. 0.78 Reka Flash 3.1
  38. 0.77 LFM2-1.2B
  39. 0.67 SmolLM2 1.7B Instruct
  40. 0.56 OLMo 2 32B Instruct
  41. 0.53 Falcon3-3B-Instruct
  42. 0.52 Gemma 3 1B
  43. 0.52 Qwen3-Next-80B-A3B
  44. 0.51 Gemma 3 1B PT
  45. 0.46 Olmo 3.1 32B Instruct
  46. 0.40 Qwen3-1.7B
  47. 0.36 SmolLM2-1.7B
  48. 0.32 Phi-2
  49. 0.32 QwQ-32B
  50. 0.32 Qwen3-4B
  51. 0.29 Gemma 3 270M
  52. 0.22 DeepSeek-R1-Distill-Qwen-32B
  53. 0.21 SmolLM2-135M
  54. 0.21 Phi-4-Reasoning-plus
  55. 0.21 LFM2.5-1.2B-Instruct
  56. 0.19 Phi-4-reasoning
  57. 0.16 DeepSeek-R1-Distill-Qwen-14B
  58. 0.13 Qwen1.5-14B-Chat
  59. 0.13 nvidia-nemotron-nano-9b-v2
  60. 0.09 SmolLM2-360M
  61. 0.09 Qwen3-0.6B
  62. 0.03 Qwen3-4B-Thinking-2507
  63. 0.0 Mistral 7B v0.3
  64. 0.0 Gemma 2 27B
  65. 0.0 DeepSeek-R1-Distill-Qwen-1.5B
  66. 0.0 DeepSeek-R1-Distill-Qwen-7B
  67. 0.0 Qwen3-8B
  68. 0.0 Phi-4-mini-reasoning
  69. 0.0 deepseek-r1-0528-qwen3-8b
  70. 0.0 GLM-4.1V-9B-Thinking
  71. 0.0 LFM2.5-1.2B-Thinking
  1. 1 Gemma 3 27B PT 16.9%
  2. 2 Apertus 70B 13.7%
  3. 3 Gemma 3 12B PT 12.9%
  4. 4 Cogito v1 Preview 70B 11.8%
  5. 5 Llama 3.1-70B 10.1%
  6. 6 Llama 4 Scout 9.84%
  7. 7 Apertus 8B Instruct 8.33%
  8. 8 Gemma 3 12B 7.11%
  9. 9 Mistral Small 3.2 7.01%
  10. 10 Mistral Small 3.1 7.00%
  11. 11 Gemma 7B 6.61%
  12. 12 Llama 3.1-8B 6.29%
  13. 13 Gemma 2 9B 6.10%
  14. 14 Gemma 3 4B PT 6.10%
  15. 15 Qwen3-Next-80B-A3B-Instruct 4.44%
  16. 16 Llama-3.1-Tulu-3-8B 4.02%
  17. 17 Gemma 3 4B 3.49%
  18. 18 Yi-1.5-34B 3.29%
  19. 19 gpt-oss-20b low 3.23%
  20. 20 Phi-4 2.56%
  21. 21 DeepSeek-R1-Distill-Llama-70B 1.97%
  22. 22 Granite-3.0-8B-Base 1.83%
  23. 23 Granite-3.0-8B-Instruct 1.80%
  24. 24 Gemma 2 2B IT 1.70%
  25. 25 Command R7B 1.53%
  26. 26 Aya Expanse 32B 1.51%
  27. 27 Gemma 2B 1.36%
  28. 28 Falcon3-10B-Instruct 1.31%
  29. 29 Qwen1.5-14B 1.22%
  30. 30 Ministral 3 3B Reasoning 0.97%
  31. 31 Reka Flash 3 0.96%
  32. 32 Llama 3.2 1B 0.95%
  33. 33 Falcon3-7B-Instruct 0.89%
  34. 34 qwen3-4b-instruct-2507 0.87%
  35. 35 Qwen3-14B 0.84%
  36. 36 AFM-4.5B 0.79%
  37. 37 Reka Flash 3.1 0.78%
  38. 38 LFM2-1.2B 0.77%
  39. 39 SmolLM2 1.7B Instruct 0.67%
  40. 40 OLMo 2 32B Instruct 0.56%
  41. 41 Falcon3-3B-Instruct 0.53%
  42. 42 Gemma 3 1B 0.52%
  43. 43 Qwen3-Next-80B-A3B thinking 0.52%
  44. 44 Gemma 3 1B PT 0.51%
  45. 45 Olmo 3.1 32B Instruct 0.46%
  46. 46 Qwen3-1.7B 0.40%
  47. 47 SmolLM2-1.7B 0.36%
  48. 48 Phi-2 0.32%
  49. 49 QwQ-32B 0.32%
  50. 50 Qwen3-4B 0.32%
  51. 51 Gemma 3 270M 0.29%
  52. 52 DeepSeek-R1-Distill-Qwen-32B 0.22%
  53. 53 SmolLM2-135M 0.21%
  54. 54 Phi-4-Reasoning-plus 0.21%
  55. 55 LFM2.5-1.2B-Instruct 0.21%
  56. 56 Phi-4-reasoning 0.19%
  57. 57 DeepSeek-R1-Distill-Qwen-14B 0.16%
  58. 58 Qwen1.5-14B-Chat 0.13%
  59. 59 nvidia-nemotron-nano-9b-v2 0.13%
  60. 60 SmolLM2-360M 0.09%
  61. 61 Qwen3-0.6B 0.09%
  62. 62 Qwen3-4B-Thinking-2507 0.03%
  63. 63 Mistral 7B v0.3 0%
  64. 64 Gemma 2 27B 0%
  65. 65 DeepSeek-R1-Distill-Qwen-1.5B 0%
  66. 66 DeepSeek-R1-Distill-Qwen-7B 0%
  67. 67 Qwen3-8B 0%
  68. 68 Phi-4-mini-reasoning 0%
  69. 69 deepseek-r1-0528-qwen3-8b 0%
  70. 70 GLM-4.1V-9B-Thinking 0%
  71. 71 LFM2.5-1.2B-Thinking 0%

Source: Official EuroEval Faroese Grammatical Error Correction leaderboard

#EntryScore ±SettingDateLinkSource
1 Gemma 3 27B PT 16.9% ±0.6 Mar 2025 hf.co Official leaderboard
2 norallm/normistral-11b-warm open weights, 11.4B 16.1% ±0.4 hf.co Official leaderboard
3 norallm/normistral-11b-long open weights, 11.4B 16% ±0.5 hf.co Official leaderboard
4 speakleash/Bielik-11B-v3-Base-20250730 open weights, 11.2B 14.2% ±0.4 hf.co Official leaderboard
5 LumiOpen/Viking-13B open weights, 14B 13.8% ±0.4 hf.co Official leaderboard
6 Apertus 70B 13.7% ±0.6 Sep 2025 hf.co Official leaderboard
7 LumiOpen/Viking-33B open weights, 33.1B 13% ±0.4 hf.co Official leaderboard
8 Gemma 3 12B PT 12.9% ±0.5 Mar 2025 hf.co Official leaderboard
9 Cogito v1 Preview 70B 11.8% ±0.3 Apr 2025 hf.co Official leaderboard
10 swiss-ai/Apertus-8B-2509 open weights, 8.1B 11.2% ±0.5 hf.co Official leaderboard
11 TrustLLMeu/baseline-7-8b_2-3t-tokens_llama open weights, 7.8B 11.1% ±0.5 hf.co Official leaderboard
12 TrustLLMeu/trustllm-v2-8b-midtrain@3c7af30bc43bb3c8607e0fab6e7f2a23dd64af49 open weights, 7.2B 11% ±0.4 hf.co Official leaderboard
13 deepcogito/cogito-v2-preview-llama-70B open weights, 70.6B 10.7% ±0.4 hf.co Official leaderboard
14 TrustLLMeu/trustllm-v2-8b@d8cc924fa550fbf9474f5251e8e2f92d1c1bf894 open weights, 7.2B 10.2% ±0.4 hf.co Official leaderboard
15 Llama 3.1-70B 10.1% ±0.6 Jul 2024 hf.co Official leaderboard
16 Llama 4 Scout 9.84% ±0.3 Apr 2025 hf.co Official leaderboard
17 LumiOpen/Llama-Poro-2-70B-SFT open weights, 70.6B 9.57% ±0.4 hf.co Official leaderboard
18 TrustLLMeu/baseline-7-8b_1t-tokens_llama open weights, 7.8B 8.53% ±0.4 hf.co Official leaderboard
19 Apertus 8B Instruct 8.33% ±0.4 Sep 2025 hf.co Official leaderboard
20 baidu/ERNIE-4.5-21B-A3B-PT open weights, 22B 8.23% ±0.5 hf.co Official leaderboard
21 Gemma 3 12B 7.11% ±0.4 Mar 2025 hf.co Official leaderboard
22 Mistral Small 3.2 7.01% ±0.5 Jun 2025 hf.co Official leaderboard
23 Mistral Small 3.1 7.00% ±0.4 Mar 2025 hf.co Official leaderboard
24 Gemma 7B 6.61% ±0.2 Feb 2024 hf.co Official leaderboard
25 Llama 3.1-8B 6.29% ±0.3 Jul 2024 hf.co Official leaderboard
26 mistralai/Mistral-Small-3.1-24B-Base-2503 open weights, 24B 6.27% ±0.5 hf.co Official leaderboard
27 Gemma 2 9B 6.10% ±0.6 Jun 2024 hf.co Official leaderboard
28 Gemma 3 4B PT 6.10% ±0.5 Mar 2025 hf.co Official leaderboard
29 deepcogito/cogito-v1-preview-llama-8B open weights, 8B 5.92% ±0.2 hf.co Official leaderboard
30 TrustLLMeu/trustllm-v2-8b-sft-8k@2aadff8653393e41abf583222164e279461a619f open weights, 7.2B 5.59% ±0.6 hf.co Official leaderboard
31 TrustLLMeu/trustllm-v2-8b-sft@f442f651172c1a97143738fc30c3a6342d1e8660 open weights, 7.2B 5.50% ±0.6 hf.co Official leaderboard
32 TrustLLMeu/trustllm-v2-8b-sft-mixb2-exp@f33c260efa853437a55d780ce5b910944b1395f6 open weights, 7.2B 5.48% ±0.4 hf.co Official leaderboard
33 norallm/normistral-11b-thinking open weights, 11.4B 5.14% ±0.3 hf.co Official leaderboard
34 allenai/Llama-3.1-Tulu-3-8B-SFT open weights, 8B 5.12% ±0.2 hf.co Official leaderboard
35 mistralai/Mistral-Small-24B-Instruct-2501 open weights, 23.6B 5.06% ±0.4 hf.co Official leaderboard
36 meta-llama/Llama-2-70b-hf open weights, 69B 5.00% ±0.2 hf.co Official leaderboard
37 mistralai/Magistral-Small-2509 open weights, 24B 5.00% ±0.4 hf.co Official leaderboard
38 deepcogito/cogito-v1-preview-qwen-32B open weights, 32.8B 4.80% ±0.3 hf.co Official leaderboard
39 Qwen/Qwen3-14B#no-thinking closed weights, 14.8B 4.72% ±0.4 hf.co Official leaderboard
40 Qwen/Qwen3-235B-A22B-Instruct-2507-FP8 closed weights, 235.1B 4.51% ±0.3 hf.co Official leaderboard
41 allenai/Llama-3.1-Tulu-3.1-8B open weights, 8B 4.44% ±0.2 hf.co Official leaderboard
42 Qwen3-Next-80B-A3B-Instruct 4.44% ±0.2 Sep 2025 hf.co Official leaderboard
43 google/gemma-2-9b-it open weights, 9.2B 4.31% ±0.2 hf.co Official leaderboard
44 LumiOpen/Llama-Poro-2-70B-Instruct open weights, 70.6B 4.30% ±0.3 hf.co Official leaderboard
45 CohereForAI/aya-23-35B open weights, 35B 4.07% ±0.3 hf.co Official leaderboard
46 NousResearch/DeepHermes-3-Llama-3-8B-Preview open weights, 8B 4.06% ±0.3 hf.co Official leaderboard
47 allenai/Llama-3.1-Tulu-3-8B-DPO open weights, 8B 4.05% ±0.3 hf.co Official leaderboard
48 Llama-3.1-Tulu-3-8B 4.02% ±0.2 Nov 2024 hf.co Official leaderboard
49 Gemma 3 4B 3.49% ±0.3 Mar 2025 hf.co Official leaderboard
50 deepcogito/cogito-v1-preview-qwen-14B open weights, 14.8B 3.29% ±0.2 hf.co Official leaderboard
51 Yi-1.5-34B 3.29% ±0.1 May 2024 hf.co Official leaderboard
52 gpt-oss-20b 3.23% ±0.2 low Aug 2025 hf.co Official leaderboard
53 allenai/OLMo-2-1124-13B open weights, 13.7B 3.17% ±0.3 hf.co Official leaderboard
54 meta-llama/Llama-2-13b-hf open weights, 13B 3.08% ±0.2 hf.co Official leaderboard
55 allenai/Olmo-3-1125-32B open weights, 32.2B 2.95% ±0.3 hf.co Official leaderboard
56 utter-project/EuroLLM-22B-2512 open weights, 22.6B 2.93% ±0.3 hf.co Official leaderboard
57 meta-llama/Llama-3.1-8B-Instruct open weights, 8B 2.92% ±0.2 hf.co Official leaderboard
58 mistralai/Mistral-Nemo-Instruct-2407 open weights, 12.3B 2.87% ±0.3 hf.co Official leaderboard
59 ai21labs/Jamba-v0.1 open weights, 51.6B 2.82% ±0.3 hf.co Official leaderboard
60 deepcogito/cogito-v1-preview-llama-3B open weights, 3.6B 2.80% ±0.1 hf.co Official leaderboard
61 utter-project/EuroLLM-22B-Instruct-2512 open weights, 22.6B 2.74% ±0.2 hf.co Official leaderboard
62 allenai/OLMo-2-0325-32B open weights, 32.2B 2.63% ±0.2 hf.co Official leaderboard
63 Qwen/Qwen3-32B#no-thinking closed weights, 32.8B 2.63% ±0.2 hf.co Official leaderboard
64 upstage/SOLAR-10.7B-v1.0 open weights, 10.7B 2.62% ±0.1 hf.co Official leaderboard
65 Phi-4 2.56% ±0.3 Dec 2024 hf.co Official leaderboard
66 01-ai/Yi-1.5-9B open weights, 8.8B 2.54% ±0.1 hf.co Official leaderboard
67 Qwen/Qwen3-30B-A3B#no-thinking closed weights, 30.5B 2.32% ±0.2 hf.co Official leaderboard
68 allenai/Olmo-3.1-32B-Instruct-SFT open weights, 32.2B 2.30% ±0.3 hf.co Official leaderboard
69 NbAiLab/borealis-4b-instruct-preview open weights, 4.3B 2.30% ±0.2 hf.co Official leaderboard
70 utter-project/EuroLLM-22B-Preview open weights 2.28% ±0.2 hf.co Official leaderboard
71 speakleash/Bielik-4.5B-v3.0-Instruct open weights, 4.8B 2.15% ±0.1 hf.co Official leaderboard
72 mistralai/Ministral-3-3B-Base-2512 open weights, 4.3B 2.11% ±0.1 hf.co Official leaderboard
73 ibm-granite/granite-20b-code-base open weights, 20.1B 2.08% ±0.1 hf.co Official leaderboard
74 mistralai/Ministral-3-8B-Reasoning-2512 open weights, 8.9B 2.06% ±0.3 hf.co Official leaderboard
75 DeepSeek-R1-Distill-Llama-70B 1.97% ±0.1 Jan 2025 hf.co Official leaderboard
76 ibm-granite/granite-8b-code-base-4k open weights, 8.1B 1.93% ±0.2 hf.co Official leaderboard
77 NovaSky-AI/Sky-T1-32B-Preview open weights, 32.8B 1.92% ±0.2 hf.co Official leaderboard
78 meta-llama/Llama-2-13b-chat-hf open weights, 13B 1.91% ±0.2 hf.co Official leaderboard
79 HuggingFaceTB/SmolLM3-3B#no-thinking open weights, 3.1B 1.90% ±0.1 hf.co Official leaderboard
80 speakleash/Bielik-11B-v2.6-Instruct open weights, 11.2B 1.90% ±0.3 hf.co Official leaderboard
81 Granite-3.0-8B-Base 1.83% ±0.2 Oct 2024 hf.co Official leaderboard
82 utter-project/EuroLLM-9B-Instruct open weights, 9.2B 1.82% ±0.2 hf.co Official leaderboard
83 Granite-3.0-8B-Instruct 1.80% ±0.2 Oct 2024 hf.co Official leaderboard
84 Qwen/Qwen3-8B#no-thinking closed weights, 8.2B 1.78% ±0.2 hf.co Official leaderboard
85 arcee-ai/Trinity-Mini-Base open weights, 26.1B 1.72% ±0.2 hf.co Official leaderboard
86 Gemma 2 2B IT 1.70% ±0.2 Jul 2024 hf.co Official leaderboard
87 ibm-granite/granite-34b-code-instruct open weights, 33.7B 1.68% ±0.2 hf.co Official leaderboard
88 utter-project/EuroLLM-9B open weights, 9.2B 1.64% ±0.2 hf.co Official leaderboard
89 Qwen/Qwen3-1.7B#no-thinking closed weights, 2B 1.63% ±0.1 hf.co Official leaderboard
90 meta-llama/Llama-3.2-3B-Instruct open weights, 3.2B 1.61% ±0.2 hf.co Official leaderboard
91 ibm-granite/granite-34b-code-base open weights, 33.7B 1.59% ±0.1 hf.co Official leaderboard
92 allenai/Olmo-3-1025-7B open weights, 7.3B 1.57% ±0.2 hf.co Official leaderboard
93 speakleash/Bielik-11B-v2.3-Instruct open weights, 11.2B 1.57% ±0.2 hf.co Official leaderboard
94 Command R7B 1.53% ±0.1 Dec 2024 hf.co Official leaderboard
95 meta-llama/Llama-2-7b-chat-hf open weights, 6.7B 1.51% ±0.1 hf.co Official leaderboard
96 Aya Expanse 32B 1.51% ±0.1 Oct 2024 hf.co Official leaderboard
97 infly/OpenCoder-8B-Instruct open weights, 7.8B 1.50% ±0.1 hf.co Official leaderboard
98 neuralmagic/Sparse-Llama-3.1-8B-2of4 open weights, 8B 1.49% ±0.2 hf.co Official leaderboard
99 tiiuae/Falcon3-10B-Base open weights, 10.3B 1.47% ±0.2 hf.co Official leaderboard
100 occiglot/occiglot-7b-eu5-instruct open weights, 7.2B 1.43% ±0.1 hf.co Official leaderboard
101 allenai/OLMo-2-1124-7B-Instruct open weights, 7.3B 1.37% ±0.2 hf.co Official leaderboard
102 Gemma 2B 1.36% ±0.2 Feb 2024 hf.co Official leaderboard
103 ibm-granite/granite-8b-code-instruct-4k open weights, 8.1B 1.35% ±0.2 hf.co Official leaderboard
104 Falcon3-10B-Instruct 1.31% ±0.1 Dec 2024 hf.co Official leaderboard
105 ibm-granite/granite-20b-code-instruct open weights, 20.1B 1.30% ±0.1 hf.co Official leaderboard
106 Qwen/Qwen3-4B#no-thinking closed weights, 4B 1.24% ±0.1 hf.co Official leaderboard
107 baidu/ERNIE-4.5-0.3B-PT open weights, 0.4B 1.23% ±0.1 hf.co Official leaderboard
108 Qwen1.5-14B 1.22% ±0.1 Feb 2024 hf.co Official leaderboard
109 Qwen/Qwen3-0.6B#no-thinking closed weights, 0.8B 1.20% ±0.1 hf.co Official leaderboard
110 utter-project/EuroLLM-22B-Instruct-Preview open weights, 22.6B 1.19% ±0.2 hf.co Official leaderboard
111 allenai/Olmo-3-7B-Instruct-SFT open weights, 7.3B 1.14% ±0.1 hf.co Official leaderboard
112 allenai/OLMo-2-1124-7B open weights, 7.3B 1.03% ±0.1 hf.co Official leaderboard
113 Qwen/Qwen2.5-Math-7B open weights, 7.6B 1.01% ±0.1 hf.co Official leaderboard
114 google/gemma-2-2b open weights, 2.6B 1.00% ±0.1 hf.co Official leaderboard
115 LiquidAI/LFM2-2.6B open weights, 2.6B 0.99% ±0.1 hf.co Official leaderboard
116 Ministral 3 3B Reasoning 0.97% ±0.1 Dec 2025 hf.co Official leaderboard
117 Reka Flash 3 0.96% ±0.1 Mar 2025 hf.co Official leaderboard
118 Llama 3.2 1B 0.95% ±0.1 Sep 2024 hf.co Official leaderboard
119 arcee-ai/AFM-4.5B-Base open weights, 4.6B 0.89% ±0.1 hf.co Official leaderboard
120 Falcon3-7B-Instruct 0.89% ±0.1 Dec 2024 hf.co Official leaderboard
121 Qwen/Qwen3-235B-A22B-Thinking-2507-FP8 closed weights, 235.1B 0.88% ±0.1 hf.co Official leaderboard
122 qwen3-4b-instruct-2507 0.87% ±0.1 Aug 2025 hf.co Official leaderboard
123 tiiuae/Falcon3-7B-Base open weights, 7.5B 0.84% ±0.1 hf.co Official leaderboard
124 Qwen3-14B 0.84% ±0.1 Apr 2025 hf.co Official leaderboard
125 common-pile/comma-v0.1-2t open weights, 7B 0.82% ±0.1 hf.co Official leaderboard
126 speakleash/Bielik-1.5B-v3.0-Instruct open weights, 1.6B 0.80% ±0.1 hf.co Official leaderboard
127 AFM-4.5B 0.79% ±0.1 Jul 2025 hf.co Official leaderboard
128 PleIAs/Pleias-1.2b-Preview open weights, 1.2B 0.78% ±0.1 hf.co Official leaderboard
129 Reka Flash 3.1 0.78% ±0.1 May 2025 hf.co Official leaderboard
130 LFM2-1.2B 0.77% ±0.1 Jul 2025 hf.co Official leaderboard
131 bineric/lynx-instruct-30b#no-thinking open weights, 30.5B 0.75% ±0.1 hf.co Official leaderboard
132 bineric/lynx-instruct-30b#thinking open weights, 30.5B 0.74% ±0.1 hf.co Official leaderboard
133 facebook/MobileLLM-R1-950M-base open weights, 0.9B 0.73% ±0.1 hf.co Official leaderboard
134 infly/OpenCoder-8B-Base open weights, 7.8B 0.73% ±0.1 hf.co Official leaderboard
135 ibm-granite/granite-3b-code-instruct-2k open weights, 3.5B 0.68% ±0.1 hf.co Official leaderboard
136 SmolLM2 1.7B Instruct 0.67% ±0.1 Oct 2024 hf.co Official leaderboard
137 ibm-granite/granite-3.0-3b-a800m-base open weights, 3.4B 0.65% ±0.1 hf.co Official leaderboard
138 meta-llama/Llama-3.2-1B-Instruct open weights, 1.2B 0.63% ±0.1 hf.co Official leaderboard
139 LiquidAI/LFM2-700M open weights, 0.7B 0.62% ±0.1 hf.co Official leaderboard
140 mii-llm/nesso-4B open weights, 4B 0.62% ±0.1 hf.co Official leaderboard
141 nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16#no-thinking open weights, 31.6B 0.62% ±0.1 hf.co Official leaderboard
142 infly/OpenCoder-1.5B-Instruct open weights, 1.9B 0.60% ±0.1 hf.co Official leaderboard
143 occiglot/occiglot-7b-eu5 open weights, 7.2B 0.57% ±0.1 hf.co Official leaderboard
144 OLMo 2 32B Instruct 0.56% ±0.1 Mar 2025 hf.co Official leaderboard
145 HuggingFaceTB/SmolLM3-3B-Base open weights, 3.1B 0.55% ±0.1 hf.co Official leaderboard
146 ibm-granite/granite-3b-code-base-2k open weights, 3.5B 0.54% ±0 hf.co Official leaderboard
147 Qwen/Qwen2.5-1.5B-Instruct open weights, 1.5B 0.54% ±0.1 hf.co Official leaderboard
148 Falcon3-3B-Instruct 0.53% ±0.1 Dec 2024 hf.co Official leaderboard
149 Gemma 3 1B 0.52% ±0.1 Mar 2025 hf.co Official leaderboard
150 Qwen3-Next-80B-A3B 0.52% ±0 thinking Sep 2025 hf.co Official leaderboard
151 HuggingFaceTB/SmolLM2-135M-Instruct open weights, 0.1B 0.51% ±0.1 hf.co Official leaderboard
152 Gemma 3 1B PT 0.51% ±0.1 Mar 2025 hf.co Official leaderboard
153 allenai/OLMo-2-1124-13B-Instruct open weights, 13.7B 0.48% ±0.1 hf.co Official leaderboard
154 Olmo 3.1 32B Instruct 0.46% ±0.1 Dec 2025 hf.co Official leaderboard
155 ibm-granite/granite-3.0-1b-a400m-instruct open weights, 1.3B 0.45% ±0 hf.co Official leaderboard
156 openGPT-X/Teuken-7B-instruct-commercial-v0.4 open weights, 7.5B 0.45% ±0.1 hf.co Official leaderboard
157 152334H/miqu-1-70b-sf open weights, 69B 0.43% ±0.1 hf.co Official leaderboard
158 common-pile/comma-v0.1-1t open weights, 7B 0.43% ±0.1 hf.co Official leaderboard
159 LiquidAI/LFM2-350M open weights, 0.4B 0.42% ±0.1 hf.co Official leaderboard
160 utter-project/EuroVLM-1.7B-Preview open weights, 2.1B 0.41% ±0.1 hf.co Official leaderboard
161 Qwen3-1.7B 0.40% ±0.1 Apr 2025 hf.co Official leaderboard
162 senseable/WestLake-7B-v2 open weights, 7.2B 0.38% ±0.1 hf.co Official leaderboard
163 SmolLM2-1.7B 0.36% ±0.1 Oct 2024 hf.co Official leaderboard
164 LiquidAI/LFM2.5-1.2B-Base open weights, 1.2B 0.35% ±0 hf.co Official leaderboard
165 allenai/Olmo-3-7B-Instruct open weights, 7.3B 0.34% ±0.1 hf.co Official leaderboard
166 allenai/Olmo-3.1-32B-Instruct-DPO open weights, 32.2B 0.34% ±0.1 hf.co Official leaderboard
167 Qwen/Qwen1.5-0.5B-Chat open weights, 0.6B 0.34% ±0.1 hf.co Official leaderboard
168 kyutai/helium-1-2b open weights, 2B 0.33% ±0.1 hf.co Official leaderboard
169 meta-llama/Llama-4-Scout-17B-16E open weights, 108.6B 0.33% ±0.2 hf.co Official leaderboard
170 utter-project/EuroMoE-2.6B-A0.6B-Instruct-Preview open weights, 2.6B 0.32% ±0.1 hf.co Official leaderboard
171 Phi-2 0.32% ±0.1 Dec 2023 hf.co Official leaderboard
172 QwQ-32B 0.32% ±0 Mar 2025 hf.co Official leaderboard
173 Qwen3-4B 0.32% ±0.1 Apr 2025 hf.co Official leaderboard
174 google/gemma-2b-it open weights, 2.5B 0.30% ±0 hf.co Official leaderboard
175 Gemma 3 270M 0.29% ±0.1 Aug 2025 hf.co Official leaderboard
176 HuggingFaceTB/SmolLM2-360M-Instruct open weights, 0.4B 0.26% ±0 hf.co Official leaderboard
177 facebook/MobileLLM-R1-360M-base open weights, 0.4B 0.25% ±0 hf.co Official leaderboard
178 meta-llama/Llama-2-70b-chat-hf open weights, 69B 0.23% ±0.1 hf.co Official leaderboard
179 allenai/Olmo-3-7B-Instruct-DPO open weights, 7.3B 0.22% ±0.1 hf.co Official leaderboard
180 ibm-granite/granite-3.0-1b-a400m-base open weights, 1.4B 0.22% ±0 hf.co Official leaderboard
181 DeepSeek-R1-Distill-Qwen-32B 0.22% ±0.1 Jan 2025 hf.co Official leaderboard
182 tiiuae/Falcon3-3B-Base open weights, 3.2B 0.21% ±0.1 hf.co Official leaderboard
183 SmolLM2-135M 0.21% ±0 Oct 2024 hf.co Official leaderboard
184 Phi-4-Reasoning-plus 0.21% ±0 Apr 2025 hf.co Official leaderboard
185 LFM2.5-1.2B-Instruct 0.21% ±0.1 Jan 2026 hf.co Official leaderboard
186 BSC-LT/salamandra-7b open weights, 7.8B 0.20% ±0.1 hf.co Official leaderboard
187 deepseek-ai/DeepSeek-R1-Distill-Llama-8B open weights, 8B 0.20% ±0 hf.co Official leaderboard
188 infly/OpenCoder-1.5B-Base open weights, 1.9B 0.20% ±0.1 hf.co Official leaderboard
189 Phi-4-reasoning 0.19% ±0.1 Apr 2025 hf.co Official leaderboard
190 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Instruct open weights, 0.1B 0.18% ±0 hf.co Official leaderboard
191 DeepSeek-R1-Distill-Qwen-14B 0.16% ±0 Jan 2025 hf.co Official leaderboard
192 HPLT/hplt2c_isl_checkpoints open weights 0.15% ±0 hf.co Official leaderboard
193 Qwen1.5-14B-Chat 0.13% ±0 Feb 2024 hf.co Official leaderboard
194 nvidia-nemotron-nano-9b-v2 0.13% ±0 Aug 2025 hf.co Official leaderboard
195 facebook/MobileLLM-R1-140M-base open weights, 0.1B 0.12% ±0 hf.co Official leaderboard
196 Qwen/Qwen1.5-0.5B open weights, 0.6B 0.11% ±0 hf.co Official leaderboard
197 HPLT/hplt2c_nob_checkpoints open weights 0.10% ±0 hf.co Official leaderboard
198 openai-community/gpt2-large open weights, 0.8B 0.10% ±0 hf.co Official leaderboard
199 BSC-LT/salamandra-7b-instruct open weights, 7.8B 0.09% ±0.1 hf.co Official leaderboard
200 SmolLM2-360M 0.09% ±0 Oct 2024 hf.co Official leaderboard
201 Qwen3-0.6B 0.09% ±0 Apr 2025 hf.co Official leaderboard
202 google/gemma-7b-it open weights, 8.5B 0.08% ±0 hf.co Official leaderboard
203 HPLT/hplt2c_dan_checkpoints open weights 0.08% ±0 hf.co Official leaderboard
204 ThatsGroes/gemma-2-27b-it-FP8-Dynamic open weights, 28.4B 0.08% ±0 hf.co Official leaderboard
205 google/gemma-2-27b-it open weights, 27.2B 0.07% ±0 hf.co Official leaderboard
206 HPLT/hplt2c_swe_checkpoints open weights 0.06% ±0 hf.co Official leaderboard
207 tiiuae/Falcon-H1-Tiny-Multilingual-100M-Base open weights, 0.1B 0.06% ±0 hf.co Official leaderboard
208 HPLT/hplt2c_nno_checkpoints open weights 0.05% ±0 hf.co Official leaderboard
209 PleIAs/Pleias-350m-Preview open weights, 0.4B 0.05% ±0 hf.co Official leaderboard
210 nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16 open weights, 4B 0.04% ±0 hf.co Official leaderboard
211 Qwen3-4B-Thinking-2507 0.03% ±0 Aug 2025 hf.co Official leaderboard
212 yulan-team/YuLan-Mini-Instruct open weights, 2.4B 0.02% ±0 hf.co Official leaderboard
213 openai-community/gpt2-medium open weights, 0.4B 0.01% ±0 hf.co Official leaderboard
214 allenai/Olmo-3-7B-Think open weights, 7.3B 0% hf.co Official leaderboard
215 BSC-LT/ALIA-40b open weights, 40.4B 0% hf.co Official leaderboard
216 DiscoResearch/DiscoLM-70b open weights, 69B 0% hf.co Official leaderboard
217 facebook/MobileLLM-R1-140M open weights, 0.1B 0% hf.co Official leaderboard
218 facebook/MobileLLM-R1-360M open weights, 0.4B 0% hf.co Official leaderboard
219 facebook/MobileLLM-R1-950M open weights, 0.9B 0% hf.co Official leaderboard
220 HuggingFaceTB/SmolLM3-3B open weights, 3.1B 0% hf.co Official leaderboard
221 MaLA-LM/emma-500-llama2-7b open weights, 6.7B 0% hf.co Official leaderboard
222 nvidia/NVIDIA-Nemotron-Nano-9B-v2#no-thinking open weights, 8.9B 0% hf.co Official leaderboard
223 openai-community/gpt2 open weights, 0.1B 0% hf.co Official leaderboard
224 PleIAs/Baguettotron@refs/pr/6#no-thinking open weights, 0.3B 0% hf.co Official leaderboard
225 PleIAs/Monad open weights, 0.1B 0% hf.co Official leaderboard
226 Qwen/Qwen2.5-Math-1.5B open weights, 1.5B 0% hf.co Official leaderboard
227 Shekswess/trlm-135m open weights, 0.1B 0% hf.co Official leaderboard
228 tencent/Hunyuan-1.8B-Instruct open weights, 1.8B 0% hf.co Official leaderboard
229 tencent/Hunyuan-7B-Instruct open weights, 7.5B 0% hf.co Official leaderboard
230 tiiuae/falcon-11B open weights, 11.1B 0% hf.co Official leaderboard
231 TildeAI/TildeOpen-30b#slow-tokenizer open weights, 30.7B 0% hf.co Official leaderboard
232 timpal0l/Mistral-7B-v0.1-flashback-v2 open weights, 7.2B 0% hf.co Official leaderboard
233 Mistral 7B v0.3 0% Sep 2023 hf.co Official leaderboard
234 Gemma 2 27B 0% Jun 2024 hf.co Official leaderboard
235 DeepSeek-R1-Distill-Qwen-1.5B 0% Jan 2025 hf.co Official leaderboard
236 DeepSeek-R1-Distill-Qwen-7B 0% Jan 2025 hf.co Official leaderboard
237 Qwen3-8B 0% Apr 2025 hf.co Official leaderboard
238 Phi-4-mini-reasoning 0% Apr 2025 hf.co Official leaderboard
239 deepseek-r1-0528-qwen3-8b 0% May 2025 hf.co Official leaderboard
240 GLM-4.1V-9B-Thinking 0% Jul 2025 hf.co Official leaderboard
241 LFM2.5-1.2B-Thinking 0% Jan 2026 hf.co Official leaderboard
#ModelScore ±SettingDateLinkSource
1 Gemma 3 27B PT 16.9% ±0.6 Mar 2025 hf.co Official leaderboard
2 Apertus 70B 13.7% ±0.6 Sep 2025 hf.co Official leaderboard
3 Gemma 3 12B PT 12.9% ±0.5 Mar 2025 hf.co Official leaderboard
4 Cogito v1 Preview 70B 11.8% ±0.3 Apr 2025 hf.co Official leaderboard
5 Llama 3.1-70B 10.1% ±0.6 Jul 2024 hf.co Official leaderboard
6 Llama 4 Scout 9.84% ±0.3 Apr 2025 hf.co Official leaderboard
7 Apertus 8B Instruct 8.33% ±0.4 Sep 2025 hf.co Official leaderboard
8 Gemma 3 12B 7.11% ±0.4 Mar 2025 hf.co Official leaderboard
9 Mistral Small 3.2 7.01% ±0.5 Jun 2025 hf.co Official leaderboard
10 Mistral Small 3.1 7.00% ±0.4 Mar 2025 hf.co Official leaderboard
11 Gemma 7B 6.61% ±0.2 Feb 2024 hf.co Official leaderboard
12 Llama 3.1-8B 6.29% ±0.3 Jul 2024 hf.co Official leaderboard
13 Gemma 2 9B 6.10% ±0.6 Jun 2024 hf.co Official leaderboard
14 Gemma 3 4B PT 6.10% ±0.5 Mar 2025 hf.co Official leaderboard
15 Qwen3-Next-80B-A3B-Instruct 4.44% ±0.2 Sep 2025 hf.co Official leaderboard
16 Llama-3.1-Tulu-3-8B 4.02% ±0.2 Nov 2024 hf.co Official leaderboard
17 Gemma 3 4B 3.49% ±0.3 Mar 2025 hf.co Official leaderboard
18 Yi-1.5-34B 3.29% ±0.1 May 2024 hf.co Official leaderboard
19 gpt-oss-20b 3.23% ±0.2 low Aug 2025 hf.co Official leaderboard
20 Phi-4 2.56% ±0.3 Dec 2024 hf.co Official leaderboard
21 DeepSeek-R1-Distill-Llama-70B 1.97% ±0.1 Jan 2025 hf.co Official leaderboard
22 Granite-3.0-8B-Base 1.83% ±0.2 Oct 2024 hf.co Official leaderboard
23 Granite-3.0-8B-Instruct 1.80% ±0.2 Oct 2024 hf.co Official leaderboard
24 Gemma 2 2B IT 1.70% ±0.2 Jul 2024 hf.co Official leaderboard
25 Command R7B 1.53% ±0.1 Dec 2024 hf.co Official leaderboard
26 Aya Expanse 32B 1.51% ±0.1 Oct 2024 hf.co Official leaderboard
27 Gemma 2B 1.36% ±0.2 Feb 2024 hf.co Official leaderboard
28 Falcon3-10B-Instruct 1.31% ±0.1 Dec 2024 hf.co Official leaderboard
29 Qwen1.5-14B 1.22% ±0.1 Feb 2024 hf.co Official leaderboard
30 Ministral 3 3B Reasoning 0.97% ±0.1 Dec 2025 hf.co Official leaderboard
31 Reka Flash 3 0.96% ±0.1 Mar 2025 hf.co Official leaderboard
32 Llama 3.2 1B 0.95% ±0.1 Sep 2024 hf.co Official leaderboard
33 Falcon3-7B-Instruct 0.89% ±0.1 Dec 2024 hf.co Official leaderboard
34 qwen3-4b-instruct-2507 0.87% ±0.1 Aug 2025 hf.co Official leaderboard
35 Qwen3-14B 0.84% ±0.1 Apr 2025 hf.co Official leaderboard
36 AFM-4.5B 0.79% ±0.1 Jul 2025 hf.co Official leaderboard
37 Reka Flash 3.1 0.78% ±0.1 May 2025 hf.co Official leaderboard
38 LFM2-1.2B 0.77% ±0.1 Jul 2025 hf.co Official leaderboard
39 SmolLM2 1.7B Instruct 0.67% ±0.1 Oct 2024 hf.co Official leaderboard
40 OLMo 2 32B Instruct 0.56% ±0.1 Mar 2025 hf.co Official leaderboard
41 Falcon3-3B-Instruct 0.53% ±0.1 Dec 2024 hf.co Official leaderboard
42 Gemma 3 1B 0.52% ±0.1 Mar 2025 hf.co Official leaderboard
43 Qwen3-Next-80B-A3B 0.52% ±0 thinking Sep 2025 hf.co Official leaderboard
44 Gemma 3 1B PT 0.51% ±0.1 Mar 2025 hf.co Official leaderboard
45 Olmo 3.1 32B Instruct 0.46% ±0.1 Dec 2025 hf.co Official leaderboard
46 Qwen3-1.7B 0.40% ±0.1 Apr 2025 hf.co Official leaderboard
47 SmolLM2-1.7B 0.36% ±0.1 Oct 2024 hf.co Official leaderboard
48 Phi-2 0.32% ±0.1 Dec 2023 hf.co Official leaderboard
49 QwQ-32B 0.32% ±0 Mar 2025 hf.co Official leaderboard
50 Qwen3-4B 0.32% ±0.1 Apr 2025 hf.co Official leaderboard
51 Gemma 3 270M 0.29% ±0.1 Aug 2025 hf.co Official leaderboard
52 DeepSeek-R1-Distill-Qwen-32B 0.22% ±0.1 Jan 2025 hf.co Official leaderboard
53 SmolLM2-135M 0.21% ±0 Oct 2024 hf.co Official leaderboard
54 Phi-4-Reasoning-plus 0.21% ±0 Apr 2025 hf.co Official leaderboard
55 LFM2.5-1.2B-Instruct 0.21% ±0.1 Jan 2026 hf.co Official leaderboard
56 Phi-4-reasoning 0.19% ±0.1 Apr 2025 hf.co Official leaderboard
57 DeepSeek-R1-Distill-Qwen-14B 0.16% ±0 Jan 2025 hf.co Official leaderboard
58 Qwen1.5-14B-Chat 0.13% ±0 Feb 2024 hf.co Official leaderboard
59 nvidia-nemotron-nano-9b-v2 0.13% ±0 Aug 2025 hf.co Official leaderboard
60 SmolLM2-360M 0.09% ±0 Oct 2024 hf.co Official leaderboard
61 Qwen3-0.6B 0.09% ±0 Apr 2025 hf.co Official leaderboard
62 Qwen3-4B-Thinking-2507 0.03% ±0 Aug 2025 hf.co Official leaderboard
63 Mistral 7B v0.3 0% Sep 2023 hf.co Official leaderboard
64 Gemma 2 27B 0% Jun 2024 hf.co Official leaderboard
65 DeepSeek-R1-Distill-Qwen-1.5B 0% Jan 2025 hf.co Official leaderboard
66 DeepSeek-R1-Distill-Qwen-7B 0% Jan 2025 hf.co Official leaderboard
67 Qwen3-8B 0% Apr 2025 hf.co Official leaderboard
68 Phi-4-mini-reasoning 0% Apr 2025 hf.co Official leaderboard
69 deepseek-r1-0528-qwen3-8b 0% May 2025 hf.co Official leaderboard
70 GLM-4.1V-9B-Thinking 0% Jul 2025 hf.co Official leaderboard
71 LFM2.5-1.2B-Thinking 0% Jan 2026 hf.co Official leaderboard

Number one over time

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

Google DeepMind
Own leaderboard Not tested since Aug 2026

Source: Official EuroEval Faroese Grammatical Error Correction 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 Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemma 3 27B PT Llama 3.1-70B Gemma 7B Qwen1.5-14B 0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemma 3 27B PT Llama 3.1-70B

Source: Official EuroEval Faroese Grammatical Error Correction 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 Multilingual

All 545 in Multilingual
  • MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
  • WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET. K-EXAONE 90.5%
  • MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language. Qwen3.7-Max 87%
  • CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories. Solar Open 2 90.7%
  • Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
  • MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 10 months ago, in January 2026, for LFM2.5-1.2B-Thinking. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (241 entries).
  • Pass Models scored 71 models we track, plus 170 other entries on its leaderboard.
  • Warning Newest score The newest came 10 months ago, in January 2026, for LFM2.5-1.2B-Thinking.
  • Note Frontier models None of the top 10 has a score. The strongest tested is QwQ-32B (#137).
  • Pass Headroom The best is 16.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on euroeval.com: 75 rows for models we track and 170 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