LLM leaderboard
Every tracked language model by intelligence, with what it costs and how fast it answers.
Claude Opus 5.5 (Max, Default Fallback) leads from Anthropic, ahead of 681 others.
- 1Anthropic
Claude Opus 5.5 (Max, Default Fallback)
57.60 intelligence
- 2Anthropic
Claude Sonnet 5.5 (Max, Default Fallback)
56.00 intelligence
- 3Anthropic
Claude Opus 5.5 (Xhigh, Default Fallback)
56.00 intelligence
ranked by intelligence · 682 models · updated Oct 7 · 1–50
| # | Model | Creator | Intelligence | $/1M in | $/1M out | tok/s |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 (Max, Default Fallback) | Anthropic | 57.60 | $4 | $20 | 97 |
| 2 | Claude Sonnet 5.5 (Max, Default Fallback) | Anthropic | 56.00 | $2 | $10 | 127 |
| 3 | Claude Opus 5.5 (Xhigh, Default Fallback) | Anthropic | 56.00 | $4 | $20 | 80 |
| 4 | Claude Opus 5.5 (High, Default Fallback) | Anthropic | 53.60 | $4 | $20 | 79 |
| 5 | Claude Fable 5.1 (Max, Default Fallback) | Anthropic | 53.40 | $10 | $50 | 65 |
| 6 | Claude Fable 5.1 (Xhigh, Default Fallback) | Anthropic | 53.20 | $10 | $50 | 60 |
| 7 | GPT-6 Astra (Max) | OpenAI | 52.70 | $10 | $50 | 59 |
| 8 | Gemini 4 Argon (High) | 52.60 | $2 | $10 | — | |
| 9 | GPT-6 Astra (Xhigh) | OpenAI | 52.40 | $10 | $50 | 59 |
| 10 | Claude Sonnet 5.5 (Xhigh, Default Fallback) | Anthropic | 51.90 | $2 | $10 | 111 |
| 11 | GPT-6.1 Sol (Max) | OpenAI | 51.80 | $2 | $10 | 60 |
| 12 | Claude Fable 5.1 (High, Default Fallback) | Anthropic | 51.20 | $10 | $50 | 58 |
| 13 | Claude Opus 5.5 (Medium, Default Fallback) | Anthropic | 51.20 | $4 | $20 | 79 |
| 14 | GPT-6.1 Sol (Xhigh) | OpenAI | 51.00 | $2 | $10 | 59 |
| 15 | GPT-6 Astra (High) | OpenAI | 50.90 | $10 | $50 | 54 |
| 16 | Claude Opus 5 (Max) | Anthropic | 50.80 | $5 | $25 | 56 |
| 17 | GPT-6.1 Sol (High) | OpenAI | 50.20 | $2 | $10 | 53 |
| 18 | Claude Opus 5 (Xhigh) | Anthropic | 49.70 | $5 | $25 | 56 |
| 19 | Claude Fable 5 (Max, Opus 4.8 Fallback) | Anthropic | 49.60 | $10 | $50 | 69 |
| 20 | GPT-6 Astra (Medium) | OpenAI | 49.60 | $10 | $50 | 50 |
| 21 | Claude Fable 5.1 (Medium, Default Fallback) | Anthropic | 48.90 | $10 | $50 | 55 |
| 22 | Claude Opus 5 (High) | Anthropic | 48.10 | $5 | $25 | 56 |
| 23 | Muse Spark 1.3 (Max) | Meta | 48.10 | $1.25 | $4.25 | 188 |
| 24 | GPT-6.1 Sol (Medium) | OpenAI | 47.80 | $2 | $10 | 50 |
| 25 | GPT-6 Sol (Max) | OpenAI | 47.60 | $2 | $10 | 95 |
| 26 | GPT-5.6 Sol (Max) | OpenAI | 47.00 | $4 | $20 | 85 |
| 27 | Claude Fable 5.1 (Low, Default Fallback) | Anthropic | 46.80 | $10 | $50 | 54 |
| 28 | Claude Sonnet 5.5 (High, Default Fallback) | Anthropic | 46.80 | $2 | $10 | 107 |
| 29 | Grok 4.7 (Xhigh) | SpaceXAI | 46.40 | $2 | $6 | 83 |
| 30 | MiMo-V2.6-Pro | Xiaomi | 46.30 | $0.435 | $0.87 | 46 |
| 31 | Grok 4.7 (High) | SpaceXAI | 46.30 | $2 | $6 | 94 |
| 32 | GPT-6 Astra (Low) | OpenAI | 45.80 | $10 | $50 | 49 |
| 33 | Qwen3.8 Max (0902) | Alibaba | 45.40 | $2 | $6 | 38 |
| 34 | Muse Spark 1.3 (Xhigh) | Meta | 45.10 | $1.25 | $4.25 | 219 |
| 35 | GLM-5.3 (Max) | Z AI | 44.80 | $1.4 | $4.4 | 77 |
| 36 | Claude Opus 5 (Medium) | Anthropic | 44.80 | $5 | $25 | 58 |
| 37 | Grok 4.6 (High) | SpaceXAI | 44.30 | $2 | $6 | 73 |
| 38 | Grok 4.6 (Xhigh) | SpaceXAI | 44.20 | $2 | $6 | 68 |
| 39 | GPT-6 Sol (Xhigh) | OpenAI | 44.20 | $2 | $10 | 93 |
| 40 | GPT-5.6 Sol (Xhigh) | OpenAI | 44.00 | $4 | $20 | 91 |
| 41 | Step 5 Preview | StepFun | 43.70 | $1 | $2.7 | 90 |
| 42 | Kimi K3 (Max) | Kimi | 43.60 | $3 | $15 | 44 |
| 43 | Grok 4.6 (Medium) | SpaceXAI | 42.80 | $2 | $6 | 71 |
| 44 | GPT-6 Sol (High) | OpenAI | 42.40 | $2 | $10 | 102 |
| 45 | Claude Opus 5.5 (Low, Default Fallback) | Anthropic | 42.30 | $4 | $20 | 80 |
| 46 | GPT-5.6 Sol (High) | OpenAI | 42.30 | $4 | $20 | 79 |
| 47 | Grok 4.7 (Low) | SpaceXAI | 42.20 | $2 | $6 | 87 |
| 48 | GPT-5.6 Terra (Max) | OpenAI | 42.10 | $2 | $12 | 118 |
| 49 | GPT-6.1 Sol (Low) | OpenAI | 42.10 | $2 | $10 | 48 |
| 50 | GLM 5.3 Flash | Z AI | 41.80 | $0.15 | $0.5 | 51 |
what intelligence means
A composite of nine independent evaluations covering reasoning, coding, agentic work and knowledge. Higher is better. It is versioned, so scores are comparable within a version rather than across all time.