OneMillion-Bench (CN): leaderboard
Metric: Pass rate (%): share of tasks whose expert score reaches the 0.7 professional threshold on the 200-task CN set (Chinese-context tasks; Finance, Law, Healthcare, Industry and Natural Science, 40 tasks each) of OneMillion-Bench's expert-curated professional tasks, vanilla models without tools (deep-research agents search by design and the paper prints the same result in both settings); rubrics graded by an LLM judge; higher is better. Source: arxiv.org. Saturation forecast: Around January 2028. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (High) | 35.5 | #76 (GPT-5.4) |
| 2 | Claude Opus 4.6 (High) | 35 | #60 (Claude Opus 4.6) |
| 3 | GPT-5.2 (High) | 22 | #105 (GPT-5.2) |
| 4 | Gemini 3 Pro (Preview) | 20 | #64 |
| 5 | Kimi K2.5 (High) | 19.5 | #139 (Kimi K2.5) |
| 6 | GPT-5.3 Codex (High) | 19 | #68 (GPT-5.3 Codex) |
| 7 | DeepSeek V3.2 Speciale | 14 | #125 |
| 8 | GLM-5 | 13 | #137 |
| 9 | Doubao-Seed-2-0-Pro-260215 (High) | 11.5 | |
| 10 | Grok 4 | 11 | #169 |
| 11 | MiniMax-M2.1 | 10 | #283 |
| 12 | MiniMax-M2.5 | 7.5 | #295 |
| 13 | Step 3.5 Flash | 6.5 | #262 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=onemillion-bench-cn · How It Works · Data refreshed daily, snapshot 2026-10-11.