LLM INDEX · Updated daily
LLM Leaderboard
Aggregated from multiple third-party benchmarks (vendor-reported scores excluded), weighted across five perspectives — coding, reasoning, office, knowledge and human blind votes. Prices are USD per million tokens; CNY is available on detail pages. Current leader: Claude Fable 5.1 (Max) (75.0 pts).
- Data snapshot
- 2026-10-22
- Ranked models
- 60
- Awaiting benchmarks
- 100
50 / 60 models · By Overall, high to low
| # | Model | Released | Overall | Coverage | |||
|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 (Max)Anthropic | 2026-09-01 | 75.0 | 86.4% 19 tests | #55 291.8B | $10.00 | $50.00 |
| 2 | GPT-6 Astra (Max)OpenAI | 2026-09-03 | 73.3 | 90.9% 20 tests | #46 419.5B | $10.00 | $50.00 |
| 3 | Claude Opus 5.5 (High)Anthropic | 2026-09-22 | 73.1 | 81.8% 18 tests | #6 7.1T | $4.00 | $20.00 |
| 4 | GPT-6.1 Sol (Max)OpenAI | 2026-09-29 | 71.1 | 81.8% 18 tests | #17 2.0T | $2.00 | $10.00 |
| 5 | Claude Opus 5 (High)Anthropic | 2026-07-24 | 70.0 | 90.9% 20 tests | #31 831.0B | $5.00 | $25.00 |
| 6 | Claude Sonnet 5.5 (xHigh)Anthropic | 2026-09-28 | 67.5 | 81.8% 18 tests | #14 2.5T | $2.00 | $10.00 |
| 7 | Gemini 4 Argon (High)Google DeepMind | 2026-09-30 | 67.3 | 50% 11 tests | #29 987.6B | $2.00 | $10.00 |
| 8 | Claude Fable 5 (High)Anthropic | 2026-06-09 | 65.8 | 50% 11 tests | — | $10.00 | $50.00 |
| 9 | GPT-6 Sol (Max)OpenAI | 2026-09-22 | 65.3 | 81.8% 18 tests | #44 433.8B | $2.00 | $10.00 |
| 10 | Grok 4.6 (High)SpaceXAI | 2026-08-12 | 64.3 | 90.9% 20 tests | #73 147.0B | $2.00 | $6.00 |
| 11 | Claude Opus 4.7Anthropic | 2026-04-16 | 64.3 | 27.3% 6 tests | — | $5.00 | $25.00 |
| 12 | GPT-5.6 Sol (xHigh)OpenAI | 2026-07-09 | 64.2 | 90.9% 20 tests | #48 407.3B | $4.00 | $20.00 |
| 13 | Gemini 3.5 Flash (Medium)Google | 2026-05-19 | 63.9 | 59.1% 13 tests | #99 69.5B | $1.50 | $9.00 |
| 14 | GPT-5.5OpenAI | 2026-04-23 | 63.8 | 95.5% 21 tests | #105 58.9B | $5.00 | $30.00 |
| 15 | Claude Opus 4.6Anthropic | 2026-02-05 | 61.5 | 27.3% 6 tests | — | $5.00 | $25.00 |
| 16 | MiMo-V2.6-ProOpenXiaomi | 2026-09-21 | 60.9 | 68.2% 15 tests | #22 1.6T | $0.43 | $0.87 |
| 17 | MiMo-V2.6-FlashOpenXiaomi | 2026-09-21 | 60.8 | 36.4% 8 tests | #4 8.6T | $0.14 | $0.28 |
| 18 | DeepSeek V4.1 Flash 1017 (max)OpenDeepSeek | 2026-10-17 | 60.8 | 45.5% 10 tests | — | $0.30 | $1.20 |
| 19 | Grok 4.7 (xHigh)SpaceXAI | 2026-09-21 | 60.7 | 81.8% 18 tests | #53 306.3B | $2.00 | $6.00 |
| 20 | GLM-5.3 FlashOpenZ AI | 2026-08-26 | 60.5 | 86.4% 19 tests | #2 11.3T | $0.01 | $0.19 |
| 21 | Muse Wonder 1.0 (max)Meta AI | 2026-10-12 | 60.5 | 36.4% 8 tests | — | — | — |
| 22 | GPT-5.6 Terra (xHigh)OpenAI | 2026-07-09 | 60.4 | 90.9% 20 tests | #59 227.9B | $2.00 | $12.00 |
| 23 | Claude Sonnet 5 (High)Anthropic | 2026-06-30 | 60.2 | 90.9% 20 tests | #37 578.2B | $2.00 | $10.00 |
| 24 | Muse Spark 1.2 (xHigh)Meta | 2026-08-05 | 59.6 | 81.8% 18 tests | #193 8.3B | $1.25 | $4.25 |
| 25 | Claude Haiku 5.5 (Low)Anthropic | 2026-10-07 | 59.5 | 77.3% 17 tests | #16 2.1T | $0.10 | $0.50 |
| 26 | Claude Opus 4.8Anthropic | 2026-05-28 | 59.1 | 50% 11 tests | — | $5.00 | $25.00 |
| 27 | GPT-5.6 Luna (xHigh)OpenAI | 2026-07-09 | 59.0 | 90.9% 20 tests | #19 1.7T | $0.20 | $1.20 |
| 28 | DeepSeek V4.1 Flash (max)OpenDeepSeek | 2026-09-09 | 58.7 | 45.5% 10 tests | #1 42.2T | $0.30 | $1.20 |
| 29 | Muse Spark 1.3 (Max)Meta AI | 2026-09-02 | 58.2 | 40.9% 9 tests | #86 96.0B | $1.25 | $4.25 |
| 30 | GPT-6 Luna (Max)OpenAI | 2026-09-22 | 57.5 | 81.8% 18 tests | #8 6.2T | $0.10 | $0.50 |
| 31 | Kimi K2.6OpenKimi | 2026-04-20 | 57.0 | 59.1% 13 tests | #77 132.9B | $0.65 | $3.41 |
| 32 | MiMo-V2.5-ProOpenXiaomi | 2026-04-22 | 56.6 | 45.5% 10 tests | #75 143.3B | $0.30 | $0.61 |
| 33 | Grok 4.5xAI | 2026-07-08 | 55.1 | 50% 11 tests | #111 45.4B | $2.00 | $6.00 |
| 34 | Gemini 3.7 Flash (High)Google DeepMind | 2026-08-13 | 54.8 | 50% 11 tests | #49 382.0B | $0.75 | $3.75 |
| 35 | Gemini 3.8 Flash (High)Google DeepMind | 2026-09-02 | 54.5 | 50% 11 tests | #20 1.6T | $0.83 | $4.13 |
| 36 | Step 5 PreviewOpenStepFun | 2026-09-18 | 54.2 | 68.2% 15 tests | #15 2.3T | $1.00 | $2.70 |
| 37 | Grok 4.20 0309 (Reasoning)SpaceXAI | 2026-03-10 | 53.6 | 18.2% 4 tests | — | — | — |
| 38 | Kimi K3 (Max)OpenMoonshot | 2026-07-16 | 52.3 | 50% 11 tests | #18 1.7T | $0.24 | $11.94 |
| 39 | Claude Sonnet 4.6Anthropic | 2026-02-17 | 51.2 | 54.5% 12 tests | — | $3.00 | $15.00 |
| 40 | Gemini 3.5 Flash-LiteGoogle | 2026-07-21 | 50.6 | 54.5% 12 tests | #60 227.4B | $0.30 | $2.50 |
| 41 | GLM-5.3 (Max)OpenZ.ai (Zhipu AI) | 2026-08-14 | 50.4 | 50% 11 tests | #9 5.2T | $0.08 | $1.40 |
| 42 | MiniMax-M3OpenMiniMax | 2026-06-01 | 48.4 | 81.8% 18 tests | #24 1.4T | $0.30 | $1.20 |
| 43 | Qwen3.8-27BOpenalibaba | — | 48.3 | 36.4% 8 tests | #45 433.6B | $0.04 | $1.35 |
| 44 | Gemini 3.1 Pro PreviewGoogle DeepMind | 2026-02-19 | 48.1 | 50% 11 tests | #78 131.0B | $2.00 | $12.00 |
| 45 | Qwen3.8-Maxalibaba | — | 47.9 | 36.4% 8 tests | #62 204.3B | — | — |
| 46 | Muse Spark Open (Xhigh)OpenMeta | 2026-10-15 | 47.5 | 59.1% 13 tests | #82 111.2B | $1.25 | $4.25 |
| 47 | InklingOpenThinking Machines | 2026-07-15 | 47.3 | 90.9% 20 tests | #56 284.0B | $1.00 | $4.05 |
| 48 | DeepSeek-V3.1-TerminusOpenDeepSeek | 2025-09-22 | 45.8 | 40.9% 9 tests | #204 5.9B | $0.27 | $1.00 |
| 49 | Mistral Large 4Mistral | 2026-10-06 | 44.6 | 68.2% 15 tests | — | — | — |
| 50 | Gemini 2.5 ProGoogle | 2025-06-05 | 42.9 | 59.1% 13 tests | #127 31.7B | $1.25 | $10.00 |
Coverage 86.4% · 19 tests · Usage #55 · Released 2026-09-01Input $10.00 · Output $50.00
Coverage 90.9% · 20 tests · Usage #46 · Released 2026-09-03Input $10.00 · Output $50.00
Coverage 81.8% · 18 tests · Usage #6 · Released 2026-09-22Input $4.00 · Output $20.00
Coverage 81.8% · 18 tests · Usage #17 · Released 2026-09-29Input $2.00 · Output $10.00
Coverage 90.9% · 20 tests · Usage #31 · Released 2026-07-24Input $5.00 · Output $25.00
Coverage 81.8% · 18 tests · Usage #14 · Released 2026-09-28Input $2.00 · Output $10.00
Coverage 50% · 11 tests · Usage #29 · Released 2026-09-30Input $2.00 · Output $10.00
Coverage 50% · 11 tests · Released 2026-06-09Input $10.00 · Output $50.00
Coverage 81.8% · 18 tests · Usage #44 · Released 2026-09-22Input $2.00 · Output $10.00
Coverage 90.9% · 20 tests · Usage #73 · Released 2026-08-12Input $2.00 · Output $6.00
Coverage 27.3% · 6 tests · Released 2026-04-16Input $5.00 · Output $25.00
Coverage 90.9% · 20 tests · Usage #48 · Released 2026-07-09Input $4.00 · Output $20.00
Coverage 59.1% · 13 tests · Usage #99 · Released 2026-05-19Input $1.50 · Output $9.00
Coverage 95.5% · 21 tests · Usage #105 · Released 2026-04-23Input $5.00 · Output $30.00
Coverage 27.3% · 6 tests · Released 2026-02-05Input $5.00 · Output $25.00
Coverage 68.2% · 15 tests · Usage #22 · Released 2026-09-21Input $0.43 · Output $0.87
Coverage 36.4% · 8 tests · Usage #4 · Released 2026-09-21Input $0.14 · Output $0.28
Coverage 45.5% · 10 tests · Released 2026-10-17Input $0.30 · Output $1.20
Coverage 81.8% · 18 tests · Usage #53 · Released 2026-09-21Input $2.00 · Output $6.00
Coverage 86.4% · 19 tests · Usage #2 · Released 2026-08-26Input $0.01 · Output $0.19
Coverage 36.4% · 8 tests · Released 2026-10-12Input — · Output —
Coverage 90.9% · 20 tests · Usage #59 · Released 2026-07-09Input $2.00 · Output $12.00
Coverage 90.9% · 20 tests · Usage #37 · Released 2026-06-30Input $2.00 · Output $10.00
Coverage 81.8% · 18 tests · Usage #193 · Released 2026-08-05Input $1.25 · Output $4.25
Coverage 77.3% · 17 tests · Usage #16 · Released 2026-10-07Input $0.10 · Output $0.50
Coverage 50% · 11 tests · Released 2026-05-28Input $5.00 · Output $25.00
Coverage 90.9% · 20 tests · Usage #19 · Released 2026-07-09Input $0.20 · Output $1.20
Coverage 45.5% · 10 tests · Usage #1 · Released 2026-09-09Input $0.30 · Output $1.20
Coverage 40.9% · 9 tests · Usage #86 · Released 2026-09-02Input $1.25 · Output $4.25
Coverage 81.8% · 18 tests · Usage #8 · Released 2026-09-22Input $0.10 · Output $0.50
Coverage 59.1% · 13 tests · Usage #77 · Released 2026-04-20Input $0.65 · Output $3.41
Coverage 45.5% · 10 tests · Usage #75 · Released 2026-04-22Input $0.30 · Output $0.61
Coverage 50% · 11 tests · Usage #111 · Released 2026-07-08Input $2.00 · Output $6.00
Coverage 50% · 11 tests · Usage #49 · Released 2026-08-13Input $0.75 · Output $3.75
Coverage 50% · 11 tests · Usage #20 · Released 2026-09-02Input $0.83 · Output $4.13
Coverage 68.2% · 15 tests · Usage #15 · Released 2026-09-18Input $1.00 · Output $2.70
Coverage 18.2% · 4 tests · Released 2026-03-10Input — · Output —
Coverage 50% · 11 tests · Usage #18 · Released 2026-07-16Input $0.24 · Output $11.94
Coverage 54.5% · 12 tests · Released 2026-02-17Input $3.00 · Output $15.00
Coverage 54.5% · 12 tests · Usage #60 · Released 2026-07-21Input $0.30 · Output $2.50
Coverage 50% · 11 tests · Usage #9 · Released 2026-08-14Input $0.08 · Output $1.40
Coverage 81.8% · 18 tests · Usage #24 · Released 2026-06-01Input $0.30 · Output $1.20
Coverage 50% · 11 tests · Usage #78 · Released 2026-02-19Input $2.00 · Output $12.00
Coverage 59.1% · 13 tests · Usage #82 · Released 2026-10-15Input $1.25 · Output $4.25
Coverage 90.9% · 20 tests · Usage #56 · Released 2026-07-15Input $1.00 · Output $4.05
Coverage 40.9% · 9 tests · Usage #204 · Released 2025-09-22Input $0.27 · Output $1.00
Coverage 68.2% · 15 tests · Released 2026-10-06Input — · Output —
Coverage 59.1% · 13 tests · Usage #127 · Released 2025-06-05Input $1.25 · Output $10.00
Showing the top 50 only (60 benchmarked models in the database; insufficient-coverage models are listed under “Awaiting benchmarks” below).
Composite scale: 50 is the reference models’ average; every 15 points is roughly one standard deviation. Coverage is the share of benchmarks available for the model. A dimension needs at least 2 contributing benchmarks to score (single-benchmark dimensions excepted); otherwise it is unscored and excluded from the composite. Switching dimensions sorts by that dimension; models without data there are hidden. Prices are USD per million tokens (converted at the day’s rate); CNY prices are on detail pages. Usage is the latest daily token volume on OpenRouter (input+output, all versions and free tiers merged); models not listed there are excluded.
Awaiting benchmarks
100 totalRecently released or below the third-party coverage bar; models enter the formal ranking automatically once they meet the threshold (≥4 aggregating benchmarks from ≥2 institutions).
Gemini 3.6 Flash (High)Google DeepMind · Human preference / Knowledge · 4 benchmarksGemini 3 ProGoogle DeepMind · Human preference · 3 benchmarksClaude Opus 4.5 (20251101)Anthropic · Human preference · 4 benchmarksGPT-5.5 InstantOpenAI · Human preference · 3 benchmarksGemini 3 Flash (Minimal)Google DeepMind · Human preference · 4 benchmarksGLM-5Z AI · Human preference · 4 benchmarksGPT-5 (High)OpenAI · Human preference / Coding · 3 benchmarksQwen3.7-Max-PreviewAlibaba · Human preference / Knowledge · 4 benchmarksKimi K2.5 (Instant)Moonshot · Human preference · 3 benchmarksClaude Opus 4.1 (20250805)Anthropic · Human preference · 3 benchmarksGPT-4.5 Preview (2025-02 27)OpenAI · Human preference · 3 benchmarksMuse-Spark-Open-816B-A42B (xhigh)Meta AI · Knowledge · 3 benchmarks
Data Sources & Methodology
Dimension weights (coding first, balanced with reasoning and office)
- Coding 40%
- Reasoning 20%
- Office 20%
- Knowledge 15%
- Human preference 5%
Methodology: each benchmark is z-score normalized across all listed models for the day, averaged per dimension, then weighted into a composite score. 50 is the reference models’ average; every 15 points is roughly one standard deviation. A model’s best reasoning tier participates in aggregation; full per-tier details are on each model page. Formal ranking requires at least 4 benchmarks from at least 2 independent evaluators; models below the coverage bar are listed as awaiting benchmarks. Updated daily.