01·Leaderboard
Every score below is at the effort you pick — not the press-release number.
Public leaderboards quote each model at maximum reasoning effort. Nobody runs xhigh for everything. Choose a tier and see what the models are actually worth there — models that don't support the tier resolve to their nearest one (shown per row).
ProprietaryOpen weights
↑ better · ← cheaperCost per typical task, log scale (1K in + 600 out + reasoning tokens)
| # | Model | Effort used | Index | GPQA | AIME | SWE-bench | Cost / task | $/M in · out | tok/s | Ctx |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5 Anthropic | High | 82.2 | 87.8 | 92.2 | 80.8 | $0.54 | $10 · $50 | 65 | 1M |
| 2 | GPT-5.5 OpenAI | High | 80.7 | 86.8 | 93.1 | 76.8 | $0.097 | $1.25 · $10 | 140 | 400K |
| 3 | Claude Opus 4.8 Anthropic | High | 79.6 | 85.8 | 89.3 | 78.8 | $0.26 | $5 · $25 | 55 | 1M |
| 4 | Gemini 3 Pro Google | High | 79.0 | 87.4 | 90.2 | 74.7 | $0.12 | $2 · $12 | 130 | 1M |
| 5 | Claude Sonnet 5 Anthropic | High | 77.2 | 84.0 | 88.0 | 76.0 | $0.12 | $3 · $15 | 80 | 1M |
| 6 | Grok 4.1 xAI | High | 76.2 | 83.5 | 87.4 | 71.8 | $0.18 | $3 · $15 | 90 | 256K |
| 7 | GLM-5.2 Zhipu AI | High | 75.8 | 81.6 | 85.6 | 74.7 | $0.021 | $0.60 · $2.20 | 95 | 200K |
| 8 | DeepSeek-V3.2 DeepSeek | High | 75.5 | 83.0 | 90.0 | 72.0 | $0.0039 | $0.28 · $0.42 | 60 | 128K |
| 9 | Kimi K2 Thinking Moonshot AI | High | 75.3 | 84.0 | 89.0 | 71.0 | $0.032 | $0.60 · $2.50 | 50 | 256K |
| 10 | Qwen3-235B-A22B Alibaba | High | 72.5 | 81.0 | 86.0 | 68.0 | $0.023 | $0.70 · $2.80 | 70 | 128K |
| 11 | Llama 4 Maverick Meta | Standard | 52.4 | 70.0 | 40.0 | 44.0 | $0.0008 | $0.27 · $0.85 | 160 | 1M |
Index = weighted composite of 8 benchmarks (weights on /benchmarks) · cost/task assumes 1K in + 600 out + the tier's typical reasoning tokens · preview data.