LLM leaderboard
Claude Fable 5 leads the LLM leaderboard with an Artificial Analysis Intelligence Index of 59.9, ahead of GPT-5.6 Sol (58.9) and Kimi K3 (57.1). 155 benchmarked models are ranked below; every row carries coding and agentic scores, live speed and price.
Every benchmarked large language model, ranked by the Artificial Analysis Intelligence Index with coding and agentic scores, live throughput and input price alongside. For image, video and voice models see the AI model leaderboard; for use-case winners see all rankings.
| # | Model | Intelligence | Coding | Agentic | Speed | In $/M |
|---|---|---|---|---|---|---|
| 1 |  claude-fable-5 | 59.9 | 76.5 | — | — | $10.00 |
| 2 |  gpt-5.6-sol | 58.9 | 77.4 | — | 74 t/s | $5.00 |
| 3 |  kimi-k3 | 57.1 | 76.2 | — | — | $3.00 |
| 4 | claude-opus-4.8 | 55.7 | 74.3 | — | — | $5.00 |
| 5 | gpt-5.6-terra | 55.0 | 76.7 | — | 96 t/s | $2.50 |
| 6 | gpt-5.5 | 54.8 | 74.9 | — | 100 t/s | $5.00 |
| 7 |  grok-4.5 | 53.8 | 72.4 | — | — | $2.00 |
| 8 | claude-opus-4.7 | 53.5 | 73.6 | — | — | $5.00 |
| 9 | claude-sonnet-5 | 53.4 | 71.5 | — | — | $2.00 |
| 10 | gpt-5.4 | 51.4 | 71.1 | — | 79 t/s | $2.50 |
| 11 | gpt-5.6-luna | 51.2 | 71.4 | — | 97 t/s | $1.00 |
| 12 |  glm-5.2 | 51.1 | 68.8 | — | — | $0.808 |
| 13 | Mmuse-spark-1.1 | 50.6 | 71.3 | — | — | $1.25 |
| 14 |  gemini-3.5-flash | 50.2 | 70.1 | — | 189 t/s | $1.50 |
| 15 | gemini-3.6-flash | 50.1 | 69.2 | — | 111 t/s | $1.50 |
| 16 | gemini-3.1-pro-preview | 46.5 | 68.8 | — | 122 t/s | $2.00 |
| 17 |  qwen3.7-max | 46.0 | 66.0 | — | — | $1.48 |
| 18 |  minimax-m3 | 44.4 | 58.6 | — | — | $0.300 |
| 19 |  deepseek-v4-pro | 44.3 | 59.4 | — | — | $0.435 |
| 20 | gpt-5.3-codex | 44.3 | — | — | — | $1.75 |
| 21 | kimi-k2.6 | 44.2 | 61.8 | — | — | $0.684 |
| 22 |  mimo-v2.5-pro | 42.2 | 60.2 | — | — | $0.435 |
| 23 | gpt-5.2 | 42.2 | — | — | 71 t/s | $1.75 |
| 24 | kimi-k2.7-code | 41.9 | 60.8 | — | — | $0.710 |
| 25 |  hy3 | 41.2 | 58.8 | — | — | $0.132 |
| 26 | Nnex-n2-pro | 41.0 | 59.1 | — | — | $0.250 |
| 27 | Tinkling | 40.7 | 52.1 | — | — | $1.00 |
| 28 | deepseek-v4-flash | 40.3 | 56.2 | — | — | $0.098 |
| 29 | glm-5.1 | 40.2 | 55.8 | — | — | $0.966 |
| 30 | gpt-5.2-codex | 40.1 | — | — | — | $1.75 |
| 31 | qwen3.6-max-preview | 40.0 | — | — | — | $1.04 |
| 32 | gpt-5.4-mini | 40.0 | 56.1 | — | 124 t/s | $0.750 |
| 33 | qwen3.6-plus | 39.6 | 54.5 | — | — | $0.325 |
| 34 | glm-5 | 39.5 | — | — | — | $0.950 |
| 35 | qwen3.7-plus | 39.0 | 55.9 | — | — | $0.320 |
| 36 | gpt-5.4-nano | 38.2 | 56.1 | — | 50 t/s | $0.200 |
| 37 | minimax-m2.7 | 38.1 | 52.6 | — | — | $0.250 |
| 38 | glm-5-turbo | 38.1 | — | — | — | $1.20 |
| 39 | claude-opus-4.6 | 37.8 | — | — | — | $5.00 |
| 40 | grok-4.3 | 37.6 | 42.2 | — | 18 t/s | $1.25 |
| 41 | mimo-v2.5 | 37.2 | 56.8 | — | — | $0.140 |
| 42 | qwen3.6-27b | 37.1 | 53.7 | — | — | $0.288 |
| 43 | grok-4.20 | 37.0 | — | — | — | $1.25 |
| 44 | gpt-5.1 | 36.9 | 49.4 | — | 78 t/s | $1.25 |
| 45 | gemini-3.5-flash-lite | 36.5 | 49.3 | — | 35 t/s | $0.300 |
| 46 | gpt-5-codex | 36.1 | — | — | — | $1.25 |
| 47 | claude-sonnet-4.6 | 35.9 | — | — | — | $3.00 |
| 48 | kimi-k2.5 | 35.4 | 46.8 | — | — | $0.570 |
| 49 | claude-opus-4.5 | 34.7 | — | — | — | $5.00 |
| 50 | gpt-5.1-codex | 34.7 | — | — | — | $1.25 |
| 51 | gpt-5 | 34.7 | 37.8 | — | — | $1.25 |
| 52 | glm-5v-turbo | 34.5 | — | — | — | $1.20 |
| 53 | qwen3.5-27b | 33.8 | — | — | — | $0.195 |
| 54 |  kat-coder-pro-v2 | 33.7 | 59.5 | — | — | $0.300 |
| 55 | qwen3.5-397b-a17b | 33.7 | 48.2 | — | — | $0.390 |
| 56 | minimax-m2.5 | 33.7 | — | — | — | $0.150 |
| 57 | glm-4.7 | 33.7 | 45.3 | — | — | $0.400 |
| 58 | hy3-preview | 33.6 | — | — | — | $0.063 |
| 59 | Mlongcat-2.0 | 33.5 | 45.3 | — | — | $0.300 |
| 60 | kimi-k2-thinking | 32.7 | — | — | — | $0.600 |
| 61 | o3-pro | 32.5 | — | — | — | $20.00 |
| 62 | qwen3.5-122b-a10b | 32.3 | 45.7 | — | — | $0.260 |
| 63 | qwen3-max-thinking | 31.7 | — | — | — | $0.780 |
| 64 | qwen3.6-35b-a3b | 31.6 | 41.9 | — | — | $0.140 |
| 65 | minimax-m2.1 | 31.4 | — | — | — | $0.300 |
| 66 | Iring-2.6-1t | 30.6 | 42.8 | — | — | $0.075 |
| 67 | gpt-5.1-codex-mini | 30.6 | — | — | — | $0.250 |
| 68 | o3 | 30.4 | — | — | — | $2.00 |
| 69 |  step-3.7-flash | 30.3 | 39.6 | — | — | $0.200 |
| 70 |  mistral-medium-3-5 | 29.9 | 46.9 | — | — | $1.50 |
| 71 | qwen3.5-35b-a3b | 29.3 | — | — | — | $0.140 |
| 72 | claude-sonnet-4.5 | 29.3 | — | — | — | $3.00 |
| 73 | minimax-m2 | 28.3 | — | — | — | $0.255 |
| 74 | claude-opus-4.1 | 28.2 | — | — | — | $15.00 |
| 75 | Iling-2.6-1t | 26.1 | — | — | — | $0.075 |
| 76 | gemini-2.5-pro | 25.8 | 33.3 | — | 85 t/s | $1.25 |
| 77 | o4-mini-high | 25.6 | — | — | — | $1.10 |
| 78 | o4-mini | 25.6 | — | — | — | $1.10 |
| 79 | step-3.5-flash | 25.5 | — | — | — | $0.100 |
| 80 | claude-opus-4 | 25.5 | — | — | — | $15.00 |
| 81 | claude-sonnet-4 | 25.5 | — | — | — | $3.00 |
| 82 | gpt-5-mini | 25.3 | 15.6 | — | 107 t/s | $0.250 |
| 83 | gemini-3.1-flash-lite | 25.0 | 34.7 | — | 109 t/s | $0.250 |
| 84 | gemini-3.1-flash-lite-preview | 25.0 | 34.7 | — | 116 t/s | $0.250 |
| 85 | deepseek-v3.2 | 24.7 | — | — | — | $0.269 |
| 86 | qwen3-max | 24.0 | — | — | — | $0.780 |
| 87 | gpt-oss-120b | 23.8 | 30.4 | — | — | $0.037 |
| 88 | claude-haiku-4.5 | 23.7 | — | — | — | $1.00 |
| 89 | kimi-k2-0905 | 23.5 | — | — | — | $0.600 |
| 90 | o1 | 23.4 | 39.7 | — | — | $15.00 |
| 91 | glm-4.6 | 23.0 | — | — | — | $0.500 |
| 92 | glm-4.7-flash | 22.9 | — | — | — | $0.060 |
| 93 |  command-a | 22.5 | 27.8 | — | — | $2.50 |
| 94 | qwen3.5-9b | 21.4 | 28.7 | — | — | $0.100 |
| 95 |  mercury-2 | 21.4 | 31.1 | — | — | $0.250 |
| 96 | deepseek-v3.1-terminus | 21.4 | — | — | — | $0.270 |
| 97 | qwen3-coder-next | 21.1 | 36.2 | — | — | $0.110 |
| 98 | deepseek-r1 | 20.1 | — | — | — | $0.700 |
| 99 | gpt-5-nano | 19.9 | — | — | 181 t/s | $0.050 |
| 100 | north-mini-code:free | 19.8 | 36.5 | — | — | Free |
Frequently asked
What is the best LLM right now?
Claude Fable 5 tops the leaderboard with a 59.9 Intelligence Index. "Best" depends on the job, though — see the coding, design, speed and price rankings for use-case winners.
How is this LLM leaderboard ranked?
Models are ordered by the Artificial Analysis Intelligence Index, an independent composite of reasoning, knowledge, math and science benchmarks (GPQA, HLE and more). Coding and Agentic index columns come from the same evaluation suite; speed and price are measured live across providers.
How is this different from Chatbot Arena (LMArena)?
Arena leaderboards rank models by crowd-voted preference; this leaderboard ranks them by independent benchmark scores, and adds the practical columns votes can't capture — tokens/sec, latency and per-token price. Design Arena's human-preference Elo is used on the design ranking.
How often does the leaderboard update?
Speed, latency and pricing refresh hourly; benchmark scores refresh daily as Artificial Analysis publishes new results. New models appear as soon as they're served by a tracked provider.