BigCodeBench
We show this table for reference; we do not rank on it.
A code-generation benchmark reported in DeepSeek-V4 base-model evaluations.
Benchmark score on BigCodeBench — November 13, 2026
We compile the BigCodeBench rows from provider self-reports. DeepSeek V4 Pro Base leads the table at 59.2%, followed by Ternary Bonsai 2 27B (58.1%) and DeepSeek V4 Flash Base (56.8%). We do not use these results to rank models overall.
DeepSeek V4 Pro Base
DeepSeek
Reported by DeepSeekTernary Bonsai 2 27B
Prism ML
Reported by Prism MLDeepSeek V4 Flash Base
DeepSeek
Reported by DeepSeek3 modelsCodingCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (3 models)
All 3 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | DeepSeek V4 Pro BaseDeepSeekReported by DeepSeek | 59.2% | Reported by DeepSeek | Not reported | Open |
| 2 | Ternary Bonsai 2 27BPrism MLReported by Prism ML | 58.1% | Reported by Prism ML | Not reported | Open |
| 3 | DeepSeek V4 Flash BaseDeepSeekReported by DeepSeek | 56.8% | Reported by DeepSeek | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported BigCodeBench rows, DeepSeek V4 Pro Base is first at 59.2%. The third row is 2.4 points behind. The broader top-10 range is 2.4 points, so many of the published results sit in a relatively narrow band.
3 models have been evaluated on BigCodeBench. The benchmark falls in the Coding category. BigCodeBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About BigCodeBench
Year
2026
Tasks
Code generation tasks
Format
Pass@1
Difficulty
Software engineering
BenchLM stores BigCodeBench as a display-only provider-table row when exact values are published in DeepSeek-V4 evaluations.
Freshness and provenance
Version
BigCodeBench 2026
Refresh cadence
Quarterly
Staleness state
Current
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 BigCodeBench measure?
A code-generation benchmark reported in DeepSeek-V4 base-model evaluations.
Which model scores highest on BigCodeBench?
DeepSeek V4 Pro Base by DeepSeek currently leads with a score of 59.2% on BigCodeBench.
How many models are evaluated on BigCodeBench?
3 AI models have published results on BigCodeBench in the BenchLM catalog.
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 7,000+ subscribers.
One email each week. Unsubscribe anytime.