DeepWeb-Bench - Derivation: leaderboard
The derivation capability family of DeepWeb-Bench (2026), a deep-research benchmark that demands massive cross-source web evidence and long-horizon derivation, scored cell by cell. The score is the mean cell score (%) over the derivation cells for agents run in Claude Code or Codex CLI with the search and browsing tools the benchmark provides; higher is better.
Metric: Mean cell score (%) over the Derivation capability family (computing derived values from the evidence), DeepWeb-Bench deep-research tasks answered cell by cell with the benchmark's web search, page visit and PDF tools (native browsing disabled), runs hosted in Claude Code CLI or Codex CLI at default settings; higher is better. Source: arxiv.org. Saturation forecast: Around May 2028. 9 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 32.55 | #54 |
| 2 | Claude Opus 4.7 | 30.97 | #85 |
| 3 | DeepSeek V4 Pro | 27.73 | #196 |
| 4 | GLM-5.1 | 27.06 | #198 |
| 5 | Claude Sonnet 4.6 | 26.87 | #173 |
| 6 | DeepSeek V4 Flash | 26.77 | #267 |
| 7 | Qwen 3.6 Plus | 25.34 | #238 |
| 8 | MiniMax-M2.7 | 22.94 | #539 |
| 9 | Kimi K2.6 | 15.36 | #192 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=deepweb-bench-derivation · How It Works · Data refreshed daily, snapshot 2026-11-21.