Kimi K3 Tech Report - Deep Research Bench: leaderboard
Moonshot AI's in-house Deep Research Bench poses deep-research queries curated by domain experts and graded with expert-aligned rubrics; distinct from other benchmarks of that name. Kimi K3 technical report, Table 3; score in percent, higher is better. Kimi K3 ran in Kimi Agent, Claude models and GLM-5.2 in Claude Code, GPT models in Codex.
Metric: Score (%). Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 (Max) | 90 | #35 (Kimi K3) |
| 2 | Claude Opus 4.8 (Max) | 87.2 | #39 (Claude Opus 4.8) |
| 3 | GPT-5.6 Sol (Max) | 85.3 | #22 (GPT-5.6 Sol) |
| 4 | GLM-5.2 (Max) | 84 | #80 (GLM-5.2) |
| 5 | GPT-5.5 (xHigh) | 81.9 | #33 (GPT-5.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kimi-k3-tech-report-deep-research-bench · How It Works · Data refreshed daily, snapshot 2026-11-05.