AGI-Eval Community Base Language Models
Base models answer Chinese and English questions across seven skill areas. The board reports objective accuracy.
The 2026/11 leaderboard reports 25 scored model entries on its primary measure.
- Made by
- AGI-Eval (Meituan)
- Released
- February 2025
- Kind
- Reasoning in many languages
- Level
- School
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- Kept private
- Measures
- Capability
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
-
DeepSeek-V3.2-Exp #99 on the capability index
- Last model added
- September 2025
- Scores refreshed
- November 21st, 2026
Kimi K2 leads with 0.71 among models released by September 2025 Kimi K2 leads with 0.71 among models released by September 2025
Best score per entry from the benchmark's own leaderboard, read on November 21st, 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 September 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 0.71 Kimi K2
- 0.68
DeepSeek-V3.1
- 0.67
DeepSeek-V3.2-Exp
- 0.67
DeepSeek-V3
- 0.66
ERNIE-4.5-300B-A47B
- 0.65 Llama4-Maverick-17B-128E-基座
- 0.63
Qwen2.5-32B
- 0.63
Qwen2.5-72B
- 0.63
Qwen3-30B-A3B
- 0.61
Qwen3-14B
- 0.61 GLM-4.5
- 0.58
Qwen3-8B
- 0.58
Llama 3.1-405B
- 0.57
Qwen2.5-14B
- 0.55
Qwen3-4B
- 0.54
Qwen2.5-7B
- 0.53
Gemma 3 27B
- 0.53
Llama 3.1-70B
- 0.52 Llama4-Scout-17B-16E-基座
- 0.51 GLM-4-9B
- 0.49
Gemma 3 12B
- 0.45
Qwen3-1.7B
- 0.36
Gemma 3 4B
- 0.36
Qwen3-0.6B
- 0.19
Gemma 3 1B
- 1 Kimi K2 Sep 2025 0.71
- 2
DeepSeek-V3.1 0.68
- 3
DeepSeek-V3.2-Exp 0.67
- 4
DeepSeek-V3 Mar 2025 0.67
- 5
ERNIE-4.5-300B-A47B 0.66
- 6 Llama4-Maverick-17B-128E-基座 0.65
- 7
Qwen2.5-32B 0.63
- 8
Qwen2.5-72B 0.63
- 9
Qwen3-30B-A3B 0.63
- 10
Qwen3-14B 0.61
- 11 GLM-4.5 0.61
- 12
Qwen3-8B 0.58
- 13
Llama 3.1-405B 0.58
- 14
Qwen2.5-14B 0.57
- 15
Qwen3-4B 0.55
- 16
Qwen2.5-7B 0.54
- 17
Gemma 3 27B 0.53
- 18
Llama 3.1-70B 0.53
- 19 Llama4-Scout-17B-16E-基座 0.52
- 20 GLM-4-9B 0.51
- 21
Gemma 3 12B 0.49
- 22
Qwen3-1.7B 0.45
- 23
Gemma 3 4B 0.36
- 24
Qwen3-0.6B 0.36
- 25
Gemma 3 1B 0.19
Source: Official AGI-Eval Community Base Language Models leaderboard
- 0.71 Kimi K2
- 0.68
DeepSeek-V3.1
- 0.67
DeepSeek-V3.2-Exp
- 0.67
DeepSeek-V3
- 0.66
ERNIE-4.5-300B-A47B
- 0.63
Qwen2.5-32B
- 0.63
Qwen2.5-72B
- 0.63
Qwen3-30B-A3B
- 0.61
Qwen3-14B
- 0.61 GLM-4.5
- 0.58
Qwen3-8B
- 0.58
Llama 3.1-405B
- 0.57
Qwen2.5-14B
- 0.55
Qwen3-4B
- 0.54
Qwen2.5-7B
- 0.53
Gemma 3 27B
- 0.53
Llama 3.1-70B
- 0.51 GLM-4-9B
- 0.49
Gemma 3 12B
- 0.45
Qwen3-1.7B
- 0.36
Gemma 3 4B
- 0.36
Qwen3-0.6B
- 0.19
Gemma 3 1B
- 1 Kimi K2 Sep 2025 0.71
- 2
DeepSeek-V3.1 0.68
- 3
DeepSeek-V3.2-Exp 0.67
- 4
DeepSeek-V3 Mar 2025 0.67
- 5
ERNIE-4.5-300B-A47B 0.66
- 6
Qwen2.5-32B 0.63
- 7
Qwen2.5-72B 0.63
- 8
Qwen3-30B-A3B 0.63
- 9
Qwen3-14B 0.61
- 10 GLM-4.5 0.61
- 11
Qwen3-8B 0.58
- 12
Llama 3.1-405B 0.58
- 13
Qwen2.5-14B 0.57
- 14
Qwen3-4B 0.55
- 15
Qwen2.5-7B 0.54
- 16
Gemma 3 27B 0.53
- 17
Llama 3.1-70B 0.53
- 18 GLM-4-9B 0.51
- 19
Gemma 3 12B 0.49
- 20
Qwen3-1.7B 0.45
- 21
Gemma 3 4B 0.36
- 22
Qwen3-0.6B 0.36
- 23
Gemma 3 1B 0.19
Source: Official AGI-Eval Community Base Language Models leaderboard
| # | Entry | Score | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | Kimi K2 Sep 2025 | 0.71 | Sep 2025 | Dec 2025 | Official leaderboard | |
| 2 | | 0.68 | Aug 2025 | Mar 2026 | Official leaderboard | |
| 3 | | 0.67 | Sep 2025 | Mar 2026 | Official leaderboard | |
| 4 | | 0.67 | Mar 2025 | Jun 2026 | Official leaderboard | |
| 5 | | 0.66 | Jun 2025 | Aug 2025 | Official leaderboard | |
| 6 | Llama4-Maverick-17B-128E-基座 | 0.65 | Mar 2026 | Official leaderboard | ||
| 7 | | 0.63 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 8 | | 0.63 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 9 | | 0.63 | Apr 2025 | Aug 2025 | Official leaderboard | |
| 10 | | 0.61 | Apr 2025 | Aug 2025 | Official leaderboard | |
| 11 | GLM-4.5 | 0.61 | Aug 2025 | Dec 2025 | Official leaderboard | |
| 12 | | 0.58 | Apr 2025 | Mar 2026 | Official leaderboard | |
| 13 | | 0.58 | Jul 2024 | Jun 2026 | Official leaderboard | |
| 14 | | 0.57 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 15 | | 0.55 | Apr 2025 | Feb 2026 | Official leaderboard | |
| 16 | | 0.54 | Sep 2024 | Mar 2026 | Official leaderboard | |
| 17 | | 0.53 | Mar 2025 | Dec 2025 | Official leaderboard | |
| 18 | | 0.53 | Jul 2024 | Dec 2025 | Official leaderboard | |
| 19 | Llama4-Scout-17B-16E-基座 | 0.52 | Feb 2026 | Official leaderboard | ||
| 20 | GLM-4-9B | 0.51 | Jun 2024 | Mar 2026 | Official leaderboard | |
| 21 | | 0.49 | Mar 2025 | Jun 2025 | Official leaderboard | |
| 22 | | 0.45 | Apr 2025 | Mar 2026 | Official leaderboard | |
| 23 | | 0.36 | Mar 2025 | Mar 2026 | Official leaderboard | |
| 24 | | 0.36 | Apr 2025 | Dec 2025 | Official leaderboard | |
| 25 | | 0.19 | Mar 2025 | Mar 2026 | Official leaderboard |
| # | Model | Score | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | Kimi K2 Sep 2025 | 0.71 | Sep 2025 | Dec 2025 | Official leaderboard | |
| 2 | | 0.68 | Aug 2025 | Mar 2026 | Official leaderboard | |
| 3 | | 0.67 | Sep 2025 | Mar 2026 | Official leaderboard | |
| 4 | | 0.67 | Mar 2025 | Jun 2026 | Official leaderboard | |
| 5 | | 0.66 | Jun 2025 | Aug 2025 | Official leaderboard | |
| 6 | | 0.63 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 7 | | 0.63 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 8 | | 0.63 | Apr 2025 | Aug 2025 | Official leaderboard | |
| 9 | | 0.61 | Apr 2025 | Aug 2025 | Official leaderboard | |
| 10 | GLM-4.5 | 0.61 | Aug 2025 | Dec 2025 | Official leaderboard | |
| 11 | | 0.58 | Apr 2025 | Mar 2026 | Official leaderboard | |
| 12 | | 0.58 | Jul 2024 | Jun 2026 | Official leaderboard | |
| 13 | | 0.57 | Sep 2024 | Jun 2026 | Official leaderboard | |
| 14 | | 0.55 | Apr 2025 | Feb 2026 | Official leaderboard | |
| 15 | | 0.54 | Sep 2024 | Mar 2026 | Official leaderboard | |
| 16 | | 0.53 | Mar 2025 | Dec 2025 | Official leaderboard | |
| 17 | | 0.53 | Jul 2024 | Dec 2025 | Official leaderboard | |
| 18 | GLM-4-9B | 0.51 | Jun 2024 | Mar 2026 | Official leaderboard | |
| 19 | | 0.49 | Mar 2025 | Jun 2025 | Official leaderboard | |
| 20 | | 0.45 | Apr 2025 | Mar 2026 | Official leaderboard | |
| 21 | | 0.36 | Mar 2025 | Mar 2026 | Official leaderboard | |
| 22 | | 0.36 | Apr 2025 | Dec 2025 | Official leaderboard | |
| 23 | | 0.19 | Mar 2025 | Mar 2026 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard in 2 groups: skill and language. Each ranks models on one part of the test. Kimi K2 leads 5 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Code reasoning
23 models we track have a score here.
- 1 Kimi K2 Sep 2025 0.52
- 2
DeepSeek-V3.1 0.51
- 3
Qwen3-30B-A3B ↑5 0.50
- 4
DeepSeek-V3.2-Exp ↓1 0.50
- 5
DeepSeek-V3 Mar 2025 ↓1 0.50
- 6
Qwen3-14B ↑3 0.48
- 7
Qwen2.5-72B 0.46
- 8
Qwen2.5-32B ↓2 0.45
- 9
Qwen3-8B ↑2 0.45
- 10
Llama 3.1-405B ↑2 0.42
Cognition
23 models we track have a score here.
- 1 Kimi K2 Sep 2025 0.86
- 2
DeepSeek-V3.1 0.86
- 3
DeepSeek-V3 Mar 2025 ↑1 0.85
- 4
DeepSeek-V3.2-Exp ↓1 0.85
- 5 GLM-4.5 ↑5 0.83
- 6
Llama 3.1-405B ↑6 0.83
- 7
Llama 3.1-70B ↑10 0.82
- 8
Qwen2.5-32B ↓2 0.82
- 9
Gemma 3 27B ↑7 0.77
- 10
Qwen2.5-72B ↓3 0.76
General reasoning
23 models we track have a score here.
- 1
ERNIE-4.5-300B-A47B ↑4 0.88
- 2 Kimi K2 Sep 2025 ↓1 0.87
- 3
DeepSeek-V3.1 ↓1 0.86
- 4
DeepSeek-V3 Mar 2025 0.85
- 5 GLM-4.5 ↑5 0.85
- 6
DeepSeek-V3.2-Exp ↓3 0.84
- 7
Qwen2.5-72B 0.82
- 8
Llama 3.1-405B ↑4 0.82
- 9
Qwen2.5-32B ↓3 0.81
- 10
Qwen3-14B ↓1 0.81
Interaction
23 models we track have a score here.
- 1
Qwen2.5-72B ↑6 0.81
- 2
Qwen2.5-32B ↑4 0.81
- 3
DeepSeek-V3.1 ↓1 0.80
- 4 Kimi K2 Sep 2025 ↓3 0.80
- 5
DeepSeek-V3 Mar 2025 ↓1 0.79
- 6
DeepSeek-V3.2-Exp ↓3 0.78
- 7
ERNIE-4.5-300B-A47B ↓2 0.78
- 8
Qwen2.5-14B ↑5 0.78
- 9
Llama 3.1-405B ↑3 0.77
- 10
Qwen3-14B ↓1 0.77
Learning
23 models we track have a score here.
- 1
DeepSeek-V3.1 ↑1 0.87
- 2
DeepSeek-V3 Mar 2025 ↑2 0.86
- 3 Kimi K2 Sep 2025 ↓2 0.86
- 4 GLM-4.5 ↑6 0.85
- 5
DeepSeek-V3.2-Exp ↓2 0.84
- 6
Llama 3.1-405B ↑6 0.80
- 7
Qwen2.5-32B ↓1 0.77
- 8
ERNIE-4.5-300B-A47B ↓3 0.76
- 9
Qwen3-4B ↑5 0.74
- 10
Qwen3-8B ↑1 0.74
Math reasoning
23 models we track have a score here.
- 1
ERNIE-4.5-300B-A47B ↑4 0.73
- 2 Kimi K2 Sep 2025 ↓1 0.70
- 3
DeepSeek-V3.2-Exp 0.62
- 4
DeepSeek-V3.1 ↓2 0.61
- 5 GLM-4.5 ↑5 0.60
- 6
Qwen3-30B-A3B ↑2 0.60
- 7
DeepSeek-V3 Mar 2025 ↓3 0.59
- 8
Qwen2.5-72B ↓1 0.58
- 9
Qwen3-14B 0.57
- 10
Qwen2.5-32B ↓4 0.57
Subject knowledge
23 models we track have a score here.
- 1 Kimi K2 Sep 2025 0.73
- 2
DeepSeek-V3.1 0.69
- 3
ERNIE-4.5-300B-A47B ↑2 0.68
- 4
DeepSeek-V3 Mar 2025 0.67
- 5
DeepSeek-V3.2-Exp ↓2 0.66
- 6 GLM-4.5 ↑4 0.66
- 7
Qwen2.5-72B 0.59
- 8
Qwen2.5-32B ↓2 0.58
- 9
Qwen3-30B-A3B ↓1 0.57
- 10
Qwen3-14B ↓1 0.55
Chinese
23 models we track have a score here.
- 1 Kimi K2 Sep 2025 0.77
- 2
ERNIE-4.5-300B-A47B ↑3 0.73
- 3
DeepSeek-V3.1 ↓1 0.72
- 4
DeepSeek-V3.2-Exp ↓1 0.71
- 5 GLM-4.5 ↑5 0.70
- 6
DeepSeek-V3 Mar 2025 ↓2 0.70
- 7
Qwen2.5-32B ↓1 0.66
- 8
Qwen2.5-72B ↓1 0.66
- 9
Qwen3-30B-A3B ↓1 0.65
- 10
Qwen3-14B ↓1 0.62
English
23 models we track have a score here.
- 1 Kimi K2 Sep 2025 0.66
- 2
DeepSeek-V3.1 0.65
- 3
DeepSeek-V3 Mar 2025 ↑1 0.64
- 4
DeepSeek-V3.2-Exp ↓1 0.63
- 5
Qwen3-30B-A3B ↑3 0.61
- 6
Qwen3-14B ↑3 0.60
- 7
Qwen2.5-72B 0.60
- 8
Qwen2.5-32B ↓2 0.59
- 9
ERNIE-4.5-300B-A47B ↓4 0.59
- 10
Qwen3-8B ↑1 0.56
4 models have held number one on the official leaderboard since its release in February 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official AGI-Eval Community Base Language Models leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AGI-Eval Community Base Language Models 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
- 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
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 14 months ago, in September 2025, for DeepSeek-V3.2-Exp. Every benchmark
Catalog record 2 pass
- Pass Task access The tasks are private.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (25 entries).
- Pass Models scored 23 models we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 14 months ago, in September 2025, for DeepSeek-V3.2-Exp.
- Note Frontier models None of the top 10 has a score. The strongest tested is DeepSeek-V3.2-Exp (#99).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on agi-eval.org: 23 rows for models we track and 2 other entries at the last read.
- Pass Last read November 21st, 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.