Benchmark
Nejumi 4: ALT_堅牢性
Higher scores are better.
Benchmark sourceScores from Nejumi Leaderboard 4 data docs.
- Models measured
- 114
- Scores recorded
- 126
- Sources
- 1
- IQ weight
- 2.00
Score distribution
- 0.826
- Mean
- 0.89
- Median
- 0.159
- Std dev
- 0.6
- 10th percentile
- 0.97
- 90th percentile
- 0.31
- Min
This benchmark has a weight of 2.00 for chat models, measured over 114 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro Previewgoogle/gemini-3.1-pro-preview· Nejumi Leaderboard 4 data docs | 1 |
| 2 | GPT-5.6 Solopenai/gpt-5.6-sol: max-effort· Nejumi Leaderboard 4 data docs | 0.995 |
| 3 | Claude Opus 5anthropic/claude-opus-5: adaptive-thinking-max· Nejumi Leaderboard 4 data docs | 0.99 |
| 4 | GPT 5.5openai/gpt-5.5-2026-04-23: xhigh-effort· Nejumi Leaderboard 4 data docs | 0.99 |
| 5 | GPT-5.6 Terraopenai/gpt-5.6-terra: max-effort· Nejumi Leaderboard 4 data docs | 0.99 |
| 6 | Gemini 2.5 Progoogle/gemini-2.5-pro· Nejumi Leaderboard 4 data docs | 0.985 |
| 7 | Claude Opus 4.7anthropic/claude-opus-4.7: adaptive-thinking-xhigh· Nejumi Leaderboard 4 data docs | 0.98 |
| 8 | Gemini 3.6 Flashgoogle/gemini-3.6-flash· Nejumi Leaderboard 4 data docs | 0.98 |
| 9 | Claude Fable 5anthropic/claude-fable-5: adaptive-thinking-max with fallback to Opus 4.8· Nejumi Leaderboard 4 data docs | 0.975 |
| 10 | Gemini 3.5 Flashgoogle/gemini-3.5-flash· Nejumi Leaderboard 4 data docs | 0.975 |
| 11 | Qwen3.5 27BQwen/Qwen3.5-27B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.975 |
| 12 | Qwen3.6 Max Previewqwen/qwen3.6-max-preview: openrouter-reasoning· Nejumi Leaderboard 4 data docs | 0.975 |
| 13 | Claude Opus 4.8anthropic/claude-opus-4.8: adaptive-thinking-xhigh· Nejumi Leaderboard 4 data docs | 0.97 |
| 14 | GLM 5.2zai-org/GLM-5.2: reasoning-max· Nejumi Leaderboard 4 data docs | 0.97 |
| 15 | GPT 5.5openai/gpt-5.5-2026-04-23: high-effort· Nejumi Leaderboard 4 data docs | 0.97 |
| 16 | GLM 5zai-org/GLM-5: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.965 |
| 17 | GPT 5.1openai/gpt-5.1-2025-11-13: high-effort· Nejumi Leaderboard 4 data docs | 0.965 |
| 18 | GPT 5.4openai/gpt-5.4-2026-03-05: high-effort· Nejumi Leaderboard 4 data docs | 0.965 |
| 19 | GPT 5.4openai/gpt-5.4-2026-03-05: xhigh-effort· Nejumi Leaderboard 4 data docs | 0.965 |
| 20 | Qwen3.5 122B-A10BQwen/Qwen3.5-122B-A10B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.965 |
| 21 | Gemini 3 Flashgoogle/gemini-3-flash-preview· Nejumi Leaderboard 4 data docs | 0.96 |
| 22 | Gemini 3 Pro Previewgoogle/gemini-3-pro-preview· Nejumi Leaderboard 4 data docs | 0.96 |
| 23 | Gemma 4 31B ITgoogle/gemma-4-31b-it· Nejumi Leaderboard 4 data docs | 0.96 |
| 24 | GPT 5.2openai/gpt-5.2-2025-12-11: xhigh-effort· Nejumi Leaderboard 4 data docs | 0.96 |
| 25 | Grok 4.5xai/grok-4.5: reasoning-high· Nejumi Leaderboard 4 data docs | 0.96 |
| 26 | Kimi K3moonshotai/kimi-k3: reasoning-max· Nejumi Leaderboard 4 data docs | 0.955 |
| 27 | o1openai/o1-2024-12-17: high-effort· Nejumi Leaderboard 4 data docs | 0.955 |
| 28 | Qwen3.5 397B A17BQwen/Qwen3.5-397B-A17B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.955 |
| 29 | Claude Opus 4.6anthropic/claude-opus-4-6: extended-thinking· Nejumi Leaderboard 4 data docs | 0.95 |
| 30 | Qwen3.5 35B A3BQwen/Qwen3.5-35B-A3B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.945 |
| 31 | Qwen3.6 27BQwen/Qwen3.6-27B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.945 |
| 32 | Claude Sonnet 4.5anthropic/claude-sonnet-4-5-20250929: extended-thinking· Nejumi Leaderboard 4 data docs | 0.94 |
| 33 | GLM 4.7zai-org/GLM-4.7: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.94 |
| 34 | Grok 4grok-4· Nejumi Leaderboard 4 data docs | 0.94 |
| 35 | GPT 5openai/gpt-5-2025-08-07: high-effort· Nejumi Leaderboard 4 data docs | 0.935 |
| 36 | Grok 4.1 Fastx-ai/grok-4-1-fast-reasoning· Nejumi Leaderboard 4 data docs | 0.935 |
| 37 | Qwen 3 235B A22B Thinking 2507Qwen/Qwen3-235B-A22B-Thinking-2507: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.935 |
| 38 | Claude Opus 4anthropic/claude-opus-4-20250514: no-thinking· Nejumi Leaderboard 4 data docs | 0.93 |
| 39 | Gemma 4 26B A4Bgoogle/gemma-4-26B-A4B-it· Nejumi Leaderboard 4 data docs | 0.93 |
| 40 | GLM 5.2z-ai/glm-5.2: openrouter-reasoning-xhigh· Nejumi Leaderboard 4 data docs | 0.93 |
| 41 | GPT-5.6 Lunaopenai/gpt-5.6-luna: max-effort· Nejumi Leaderboard 4 data docs | 0.93 |
| 42 | Qwen3.5 9BQwen/Qwen3.5-9B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.93 |
| 43 | GLM 4.6zai-org/GLM-4.6-FP8: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.925 |
| 44 | OpenAI o3openai/o3-2025-04-16: high-effort· Nejumi Leaderboard 4 data docs | 0.925 |
| 45 | Claude Opus 4.1anthropic/claude-opus-4-1-20250805: extended-thinking· Nejumi Leaderboard 4 data docs | 0.92 |
| 46 | Grok 4.20xai/grok-4.20-0309-reasoning· Nejumi Leaderboard 4 data docs | 0.92 |
| 47 | Qwen3 VL 32B ThinkingQwen/Qwen3-VL-32B-Thinking· Nejumi Leaderboard 4 data docs | 0.92 |
| 48 | Qwen3.6 35B A3BQwen/Qwen3.6-35B-A3B: reasoning-enabled· Nejumi Leaderboard 4 data docs | 0.92 |
| 49 | Claude Sonnet 4.6anthropic/claude-sonnet-4.6: extended-thinking· Nejumi Leaderboard 4 data docs | 0.91 |
| 50 | DeepSeek V4 Flashdeepseek-ai/DeepSeek-V4-Flash: reasoning-max· Nejumi Leaderboard 4 data docs | 0.91 |
Showing 1–50 of 126 results.
Data credits: Nejumi Leaderboard 4, Weights & Biases Japan; public publisher-selected numeric evaluation results