AI Model Benchmarks: Leaderboards

The RuntimeWire Composite Index averages each model's published leaderboard scores into one 0–100 ranking, alongside N-way benchmarks where 2–30 models run the same task suite, scored objectively or by a blind LLM judge.

RuntimeWire Composite Index

Incorporates 4 evaluations (BBEH Mini, Editorial Craft, Mazur Constrained Story Writing (50-task sample), Newsroom Reliability v0.2) — the equal-weight average of each model's published leaderboard scores on a 0–100 scale.

  1. Anthropic: Claude Opus 5.5 — 99.9 (covers 1/4 evaluations)
  2. OpenAI: GPT-6 Astra — 99.5 (covers 1/4 evaluations)
  3. OpenAI: GPT-6.1 Sol Pro — 99.4 (covers 1/4 evaluations)
  4. Anthropic: Claude Sonnet 5.5 — 97.6 (covers 1/4 evaluations)
  5. Z.ai: GLM 5.3 Prime — 97.1 (covers 1/4 evaluations)
  6. xAI: Grok Latest — 83.8 (covers 1/4 evaluations)
  7. Z.ai: GLM 5.3 Flash — 81.9 (covers 2/4 evaluations)
  8. Claude Opus 5 (Fast) — 80.5 (covers 2/4 evaluations)
  9. Qwen: Qwen3.8 Flash — 79.6 (covers 2/4 evaluations)
  10. Google: Gemini 3.7 Flash — 78.1 (covers 2/4 evaluations)
  11. step-3.7-flash — 75.3 (covers 2/4 evaluations)
  12. DeepSeek-V4-Flash-0731 — 71.7 (covers 1/4 evaluations)
  13. Google: Gemini 3.8 Flash — 69.9 (covers 1/4 evaluations)
  14. SpaceXAI: Grok 4.6 — 66.5 (covers 1/4 evaluations)
  15. Qwen: Qwen3.8 Max Prime — 66.4 (covers 1/4 evaluations)