LLM Leaderboard
Ranked by GPQA Diamond · source AA
Graduate-level science Q&A. Strong reasoning signal.
#1 by GPQA Diamond
GPT-6 Astra (Xhigh)
96.3%
OpenAI · in $10.0 / out $50.0 per 1M · 52 tok/s
| # | Model | GPQA |
|---|---|---|
| 1 | GPT-6 Astra (Xhigh) | 96.3% |
| 2 | GPT-6 Astra (Max) | 96.1% |
| 3 | Gemini 3.8 Flash (High) | 95.3% |
| 4 | GPT-6 Astra (High) | 94.9% |
| 5 | Grok 4.6 (High) | 94.9% |
| 6 | Gemini 3.7 Flash (High) | 94.5% |
| 7 | Muse Spark 1.3 (Xhigh) | 94.1% |
| 8 | Gemini 3.1 Pro Preview | 94.1% |
| 9 | GPT-5.6 Sol (Max) | 94.1% |
| 10 | GPT-6 Astra (Medium) | 93.9% |
| 11 | Muse Spark 1.3 (max) | 93.8% |
| 12 | Claude Fable 5.1 (Max, Default Fallback) | 93.7% |
| 13 | Claude Opus 5 (Xhigh) | 93.7% |
| 14 | Claude Opus 5 (High) | 93.7% |
| 15 | Muse Spark 1.3 (Max) | 93.5% |
| 16 | Gemini 3.8 Flash (Medium) | 93.5% |
| 17 | Grok 4.6 (Xhigh) | 93.5% |
| 18 | Grok 4.6 (Medium) | 93.5% |
| 19 | Kimi K3 (Max) | 93.5% |
| 20 | GPT-5.5 (Xhigh) | 93.5% |
| 21 | Qwen3.8 2.4T A95B | 93.5% |
| 22 | Claude Fable 5.1 (Xhigh, Default Fallback) | 93.4% |
| 23 | GPT-5.5 (High) | 93.2% |
| 24 | Claude Opus 5 (Max) | 93.2% |
| 25 | GPT-6 Astra (Low) | 93.1% |
| 26 | Grok 4.5 (High) | 93.1% |
| 27 | GPT-5.6 Sol (Xhigh) | 93.1% |
| 28 | MiniMax-M3 | 92.9% |
| 29 | Qwen3.8 Max (0902) | 92.8% |
| 30 | DeepSeek V4 Pro 0813 (Max) | 92.8% |
| 31 | Gemini 3.6 Flash (High) | 92.8% |
| 32 | GPT-5.6 Sol (High) | 92.8% |
| 33 | Qwen3.8 Max | 92.7% |
| 34 | Claude Fable 5 (Max, Opus 4.8 Fallback) | 92.6% |
| 35 | GPT-5.5 (Medium) | 92.6% |
| 36 | GPT-5.6 Sol (Medium) | 92.6% |
| 37 | GPT-5.6 Terra (Max) | 92.5% |
| 38 | Agnes 3.0 Flash | 92.4% |
| 39 | Qwen3.7 Max | 92.3% |
| 40 | Qwen3.8-Flash-Next | 92.3% |
| 41 | Gemini 3.5 Flash (High) | 92.2% |
| 42 | Gemini 3.7 Flash (Medium) | 92.1% |
| 43 | Gemini 3.5 Flash (Medium) | 92.1% |
| 44 | Gemini 3.8 Flash (Low) | 92.0% |
| 45 | Claude Opus 4.8 (Max) | 92.0% |
| 46 | GPT-5.4 (Xhigh) | 92.0% |
| 47 | Claude Opus 5 (Medium) | 91.9% |
| 48 | GLM-5.3 (Max) | 91.7% |
| 49 | GPT-5.3 Codex (Xhigh) | 91.5% |
| 50 | Claude Opus 4.7 (Max) | 91.4% |
Quality, pricing, and speed benchmarks via Artificial Analysis. Human preference Elo via LMArena. Refreshed daily at 06:00 UTC · last sync .
About the metrics
Intelligence Index
AA
Artificial Analysis composite of reasoning, knowledge, coding, math.
LMArena Elo
LMArena
Human preference Elo from chatbot arena head-to-head votes.
Coding Index
AA
AA composite — LiveCodeBench, SciCode, etc.
Math Index
AA
AA composite — MATH-500, AIME, etc.
MMLU-Pro
AA
Multi-domain knowledge & reasoning. Harder than MMLU.
GPQA Diamond
AA
Graduate-level science Q&A. Strong reasoning signal.
LiveCodeBench
AA
Competitive programming, contamination-resistant.
MATH-500
AA
Subset of the MATH benchmark.
AIME
AA
American Invitational Mathematics Examination — olympiad math.
Humanity's Last Exam
AA
Frontier reasoning benchmark; most models still score low.
SciCode
AA
Scientific coding tasks.