reasoning
Humanity's Last Exam
- #191default of 58722.6%
Leader: Opus 5.5 · 61.4%
Ant Group · released 2026-09-11
Measured on 19 benchmarks.
General frontier
24.3
#96 of 136 models
includes estimates: 24% of the mix is measured for this model
reasoning
Leader: Opus 5.5 · 61.4%
knowledge
Leader: Opus 5.5 · 46.4
agentic
Leader: Gemini 3.7 Flash · 60.0%
agentic
Leader: Sonnet 5.5 · 1823
agentic
Leader: Gemini 3.5 Flash · 47.1%
agentic
Leader: Gemini 4 · 77.5%
agentic
Leader: Qwen 3.8 Max · 51.3%
coding
Leader: Opus 5.5 · 66.9%
coding
Leader: Fable 5.1 · 91.4%
coding
Leader: Sonnet 5.5 · 63.6%
composite
Leader: Opus 5.5 · 57.6
economics
Leader: Muse Wonder Full (Max) · 32.6%
economics
Leader: Opus 5.5 · 1866
knowledge
Leader: Fable 5.1 · 67.2%
knowledge
Leader: MiniCPM5-1B (Non-reasoning) · 99.1%
long-context
Leader: Kimi K3 · 88.7%
reasoning
Leader: Muse Wonder Full (Max) · 32.9%
speed
Leader: Gemini 2.5 Flash-Lite · 0.3s
speed
Leader: Celeris-1 · 1522.5