Inference

Humanity's Last Exam with tools

Tool-augmented Humanity's Last Exam scores reported in DeepSeek-V4 thinking-mode evaluations.

  • In index
  • Agentic
ModelsModels with a result.20
Top resultBest result on this test.67.7%Claude Opus 5.5
Top-3 spreadPoints from first to third.3.8 pts
YearYear of release.N/A

Result and price

020406080100$0.1$0.3$1$3$10$30$100RESULTOUTPUT PRICE PER 1M TOKENS · LOG SCALE

Results

20 results
#ModelResultThe score from the source.
01Claude Opus 5.567.7%
02Claude Opus 564.7%
03DeepSeek V4.1 Flash63.9%
04GLM-5.362.5%
05DeepSeek V4 Pro 081360.0%
06Claude Sonnet 557.4%
07GPT-6 Astra57.2%
08Qwen3.8 Max56.2%
09Apodex 1.156.1%
09Ornith-1.5-397B56.1%
11Hy4 preview55.4%
12GLM-5.3-Flash55.3%
13Qwen3.7 Max53.5%
14dots3-note Preview52.6%
15Agents-A147.6%
16Step 3.7 Flash47.2%
17DeepSeek V4 Flash 073145.1%
18Nemotron 3 Ultra37.4%
19Ornith-1.5-35B-A3B33.4%
20Ornith-1.5-9B30.5%