HumanEval
HumanEval remains the most-cited code-generation benchmark despite frontier models saturating it (>95% on every flagship). Kept in the index as a baseline filter — any serious model has to clear ~90% — and because vendor cards still report it. For real-world signal, weight Aider Polyglot more heavily.
Category: coding · Source: github.com ↗
| # | Model | Score | Setting | Measured | Source |
|---|---|---|---|---|---|
| 01 |  Claude 4.7 OpusAnthropic | 97.4% | pass@1 | Apr 16, 2026 | link ↗ |
| 02 |  GPT-5.5OpenAI | 96.8% | pass@1 | Apr 23, 2026 | link ↗ |
| 03 |  Gemini 3 Pro | 96.5% | pass@1 | Apr 22, 2026 | link ↗ |
| 04 | Claude 4.6 Sonnet Anthropic | 95.9% | pass@1 | Feb 17, 2026 | link ↗ |
| 05 | GPT-5 OpenAI | 95.2% | pass@1 | Aug 7, 2025 | link ↗ |
| 06 |  DeepSeek-V4-ProDeepSeek | 95.1% | pass@1 | Apr 22, 2026 | link ↗ |
| 07 |  Qwen3.7-MaxAlibaba | 93.9% | pass@1 | May 20, 2026 | link ↗ |
| 08 | Qwen3.6-27B Alibaba | 91.5% | pass@1 | Apr 22, 2026 | link ↗ |
| 09 |  Mistral Large 2Mistral | 89.0% | pass@1 | Jul 24, 2024 | link ↗ |
Other model leaderboards
- gpt.buzz Composite Model Index — cross-benchmark ranking
- MMLU-Pro — Multitask language understanding across 14 disciplines — the harder, contamination-resistant successor to MMLU. 10-option multiple choice with stronger distractors.
- GPQA Diamond — PhD-level science questions in biology, physics, and chemistry, written by domain experts. The hardest subset of GPQA — humans with PhDs in the subject score ~65%.
- Aider Polyglot — Multi-language code edit benchmark — solve real Exercism problems across Python, JavaScript, Rust, Go, Java, C++ via edit-and-test loops.
- AIME 2025 — American Invitational Mathematics Examination — competition math problems requiring multi-step reasoning. Reasoning-tier models score in the 90s; non-reasoning in the 50s.
- LiveCodeBench — Contamination-free code benchmark — only problems published AFTER a model's training cutoff. Refreshed monthly.
- MMMU — Massive Multi-discipline Multimodal Understanding — 11.5k college-level multimodal questions across 30 subjects (art, business, science, medicine, etc.).