LLM Leaderboard
Ranked by Humanity's Last Exam · source AA
Frontier reasoning benchmark; most models still score low.
#1 by Humanity's Last Exam
Claude Opus 5.5 (Max, Default Fallback)
61.4%
Anthropic · in $4.00 / out $20.0 per 1M · 96 tok/s
| # | Model | HLE |
|---|---|---|
| 1 | Claude Opus 5.5 (Max, Default Fallback) | 61.4% |
| 2 | Claude Fable 5.1 (Max, Default Fallback) | 59.1% |
| 3 | Claude Fable 5.1 (Xhigh, Default Fallback) | 58.7% |
| 4 | Claude Opus 5.5 (Xhigh, Default Fallback) | 57.5% |
| 5 | Gemini 4 Argon (High) | 57.1% |
| 6 | Claude Fable 5.1 (High, Default Fallback) | 55.9% |
| 7 | Claude Opus 5.5 (High, Default Fallback) | 55.6% |
| 8 | Claude Fable 5 (Max, Opus 4.8 Fallback) | 55.5% |
| 9 | Claude Sonnet 5.5 (Max, Default Fallback) | 55.0% |
| 10 | Claude Opus 5 (Max) | 54.9% |
| 11 | GPT-6 Astra (Max) | 54.7% |
| 12 | Claude Opus 5.5 (Medium, Default Fallback) | 54.7% |
| 13 | GPT-6 Astra (Xhigh) | 54.6% |
| 14 | Claude Opus 5 (Xhigh) | 54.4% |
| 15 | Claude Fable 5.1 (Medium, Default Fallback) | 53.8% |
| 16 | Muse Wonder Full (Max) | 53.6% |
| 17 | GPT-6 Astra (High) | 53.1% |
| 18 | GPT-6.1 Sol (Max) | 52.9% |
| 19 | Claude Opus 5 (High) | 52.8% |
| 20 | GPT-6 Astra (Medium) | 52.7% |
| 21 | GPT-6.1 Sol (Xhigh) | 52.6% |
| 22 | Kimi K3.1 Pro | 52.5% |
| 23 | GPT-6.1 Sol (High) | 51.4% |
| 24 | Claude Opus 5 (Medium) | 51.3% |
| 25 | Muse Wonder (Max) | 50.1% |
| 26 | Claude Sonnet 5.5 (Xhigh, Default Fallback) | 50.0% |
| 27 | GPT-6.1 Sol (Medium) | 49.9% |
| 28 | GPT-5.6 Sol (Max) | 49.5% |
| 29 | MiMo-V2.6-Pro | 49.4% |
| 30 | GPT-6 Astra (Low) | 49.2% |
| 31 | Muse Spark 1.3 (max) | 49.1% |
| 32 | Claude Fable 5.1 (Low, Default Fallback) | 48.9% |
| 33 | Muse Spark 1.3 (Max) | 48.7% |
| 34 | Claude Opus 4.8 (Max) | 48.7% |
| 35 | Claude Opus 5.5 (Low, Default Fallback) | 48.3% |
| 36 | MiniMax-M3.1-Flash | 48.0% |
| 37 | GPT-6 Sol (Max) | 47.9% |
| 38 | Gemini 3.7 Flash (High) | 47.9% |
| 39 | Gemini 3.8 Flash (High) | 47.8% |
| 40 | Muse Spark 1.3 (Xhigh) | 47.5% |
| 41 | GPT-6.1 Sol (Low) | 47.4% |
| 42 | GPT-5.6 Sol (Xhigh) | 47.3% |
| 43 | Gemini 3.1 Pro Preview | 47.0% |
| 44 | Kimi K3 (Max) | 46.9% |
| 45 | Step 5 Preview | 46.5% |
| 46 | GPT-6 Sol (Xhigh) | 46.3% |
| 47 | GPT-6.1 Terra (Max) | 46.2% |
| 48 | Muse Spark 1.1 (Xhigh) | 46.2% |
| 49 | GPT-5.6 Sol (High) | 46.0% |
| 50 | Claude Sonnet 5.5 (High, Default Fallback) | 45.8% |
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.