LLM Leaderboard
Ranked by SciCode · source AA
Scientific coding tasks.
#1 by SciCode
Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)
60.2%
Anthropic · in $10.0 / out $50.0 per 1M · 57 tok/s
| # | Model | SciCode |
|---|---|---|
| 1 | Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | 60.2% |
| 2 | Gemini 3.1 Pro Preview | 58.9% |
| 3 | Kimi K3 | 58.7% |
| 4 | Muse Spark 1.1 (xhigh) | 58.2% |
| 5 | GPT-5.6 Sol (high) | 56.9% |
| 6 | GPT-5.4 (xhigh) | 56.6% |
| 7 | GPT-5.6 Sol (medium) | 56.5% |
| 8 | GPT-5.6 Sol (max) | 56.1% |
| 9 | GPT-5.5 (xhigh) | 56.1% |
| 10 | Gemini 3 Pro Preview (high) | 56.1% |
| 11 | GPT-5.6 Sol (xhigh) | 56.0% |
| 12 | GPT-5.5 (high) | 55.9% |
| 13 | GPT-5.6 Sol (low) | 55.4% |
| 14 | GPT-5.2 Codex (xhigh) | 54.6% |
| 15 | Claude Opus 4.7 (Adaptive Reasoning, Max Effort) | 54.5% |
| 16 | Grok 4.5 (high) | 54.1% |
| 17 | GPT-5.6 Terra (max) | 53.9% |
| 18 | Claude Sonnet 5 (Adaptive Reasoning, Max Effort) | 53.6% |
| 19 | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | 53.5% |
| 20 | GPT-5.5 (medium) | 53.5% |
| 21 | Kimi K2.6 | 53.5% |
| 22 | GPT-5.3 Codex (xhigh) | 53.2% |
| 23 | Gemini 3.5 Flash (high) | 53.1% |
| 24 | Gemini 3.5 Flash (medium) | 53.0% |
| 25 | GPT-5.6 Luna (max) | 52.5% |
| 26 | GPT-5.2 (xhigh) | 52.1% |
| 27 | Claude Opus 4.6 (Adaptive Reasoning, Max Effort) | 51.9% |
| 28 | GPT-5.6 Terra (xhigh) | 51.6% |
| 29 | GPT-5.5 (low) | 51.6% |
| 30 | Muse Spark | 51.5% |
| 31 | GPT-5.6 Luna (high) | 50.7% |
| 32 | Gemini 3 Flash Preview (Reasoning) | 50.6% |
| 33 | GLM-5.2 (max) | 50.5% |
| 34 | GPT-5.4 (low) | 50.3% |
| 35 | GPT-5.5 Instant (May 2026) | 50.3% |
| 36 | MiMo-V2.5-Pro | 50.2% |
| 37 | Grok Build 0.1 0616 | 50.2% |
| 38 | GPT-5.6 Terra (high) | 50.1% |
| 39 | Claude Opus 4.7 (Non-reasoning, High Effort) | 50.1% |
| 40 | GPT-5.6 Luna (xhigh) | 50.0% |
| 41 | DeepSeek V4 Pro (Reasoning, Max Effort) | 50.0% |
| 42 | GPT-5.4 mini (xhigh) | 49.9% |
| 43 | Gemini 3 Pro Preview (low) | 49.9% |
| 44 | Gemini 3 Flash Preview (Non-reasoning) | 49.9% |
| 45 | GPT-5.6 Terra (medium) | 49.7% |
| 46 | Claude Opus 4.5 (Reasoning) | 49.5% |
| 47 | GPT-5.6 Terra (low) | 49.2% |
| 48 | Kimi K2.5 (Reasoning) | 49.0% |
| 49 | Qwen3.7 Max | 48.8% |
| 50 | Gemini 3.5 Flash (minimal) | 48.8% |
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.