C-Eval
We show this table for reference; we do not rank on it.
A Chinese-language academic and professional benchmark spanning humanities, social science, STEM, and applied subjects.
Benchmark score on C-Eval — October 29, 2026
We compile the C-Eval rows from provider self-reports and secondary reports. Kimi K2.5 leads the table at 94%, followed by Qwen3.6 Plus (93.3%) and DeepSeek V4 Pro Base (93.1%). We do not use these results to rank models overall.
Kimi K2.5
Moonshot AI
Qwen3.6 Plus
Alibaba
DeepSeek V4 Pro Base
DeepSeek
14 modelsKnowledgeStaleDisplay onlyUpdated October 29, 2026
Benchmark score table (14 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Kimi K2.5Moonshot AI | 94% | Not reported | Open |
| 2 | Qwen3.6 PlusAlibaba | 93.3% | Not reported | Closed |
| 3 | DeepSeek V4 Pro BaseDeepSeek | 93.1% | Not reported | Open |
| 4 | Qwen3.5 397BAlibaba | 93% | Not reported | Open |
| 5 | Claude Opus 4.5Anthropic | 92.2% | Not reported | Closed |
| 6 | DeepSeek V4 Flash BaseDeepSeek | 92.1% | Not reported | Open |
| 7 | Qwen3.6-27BAlibaba | 91.4% | Not reported | Open |
| 8 | Qwen3.6-35B-A3BAlibaba | 90% | Not reported | Open |
| 9 | Qwen3 235B-A22B (Reasoning)Alibaba | 89.6% | Not reported | Open |
| 10 | QwQ-32BAlibaba | 88.4% | Not reported | Open |
| 11 | Qwen3 32B (Reasoning)Alibaba | 87.3% | Not reported | Open |
| 12 | Qwen3 235B-A22BAlibaba | 86.1% | Not reported | Open |
| 13 | LongCat-Flash-Lite-SparseMeituan | 85.8% | Not reported | Open |
| 14 | Qwen3 32BAlibaba | 83.3% | Not reported | Open |
Among the reported C-Eval rows, Kimi K2.5 is first at 94%. The third row is 0.9 points behind. The broader top-10 range is 5.6 points, so many of the published results sit in a relatively narrow band.
14 models have been evaluated on C-Eval. The benchmark falls in the Knowledge category. C-Eval is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About C-Eval
Year
2023
Tasks
Chinese academic and professional exams
Format
Multiple choice questions
Difficulty
High school to professional level
C-Eval is one of the clearest public signals for non-English academic knowledge performance. It tests whether a model can sustain strong factual recall and reasoning under Chinese-language exam conditions across many domains.
Freshness and provenance
Version
C-Eval 2023
Refresh cadence
Static
Staleness state
Stale
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does C-Eval measure?
A Chinese-language academic and professional benchmark spanning humanities, social science, STEM, and applied subjects.
Which model scores highest on C-Eval?
Kimi K2.5 by Moonshot AI currently leads with a score of 94% on C-Eval.
How many models are evaluated on C-Eval?
14 AI models have published results on C-Eval in the BenchLM catalog.
Compare top models on C-Eval
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.