DeepSeek·released 2023-11-021 source
DeepSeek Coder 33B benchmark scores: 4 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 4 benchmarks·best result 35.4 on GSM8K (self-reported)·0 of 4 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| GSM8K | 35.4accuracy (%) | self-reported· optimizedT1 | 2023-11-02Qwen2.5-Coder Technical Report |
| Winogrande | 24accuracy (%) | self-reported· optimizedT1 | 2023-11-02Qwen2.5-Coder Technical Report |
| ARC AI2 | 22.93accuracy (%) | self-reported· optimizedT1 | 2023-11-02Qwen2.5-Coder Technical Report |
| MMLU | 19.2accuracy (%) | self-reported· optimizedT1 | 2023-11-02Qwen2.5-Coder Technical Report |
Claims drawn from cited facts, not live model generation.
This model was developed by DeepSeek and released in 2023. It is a large 33-billion-parameter model, a scale that offers substantial headroom for complex tasks but also brings higher computational expense, whereas smaller models would be cheaper and more efficient.
3 cited facts
This model is scored on 4 tracked benchmarks. It currently holds no top score on any of them. The scores are self-reported and not independently confirmed, so treat them as claims rather than verified results.
3 cited facts
With an average gap of 61.68 points behind the leader, this model trails well behind the front-of-pack. It holds the top score on none of the benchmarks (0 of them), meaning it has no genuine category leadership yet.
2 cited facts
2 facts cross-checked across data sources: 2 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: curated_param_counts, epoch_benchmarks
DeepSeek Coder 33B is an AI model developed by DeepSeek, released 2023-11-02. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
DeepSeek Coder 33B has recorded scores on 4 benchmarks, each shown with its evidence status.
DeepSeek Coder 33B has recorded scores on 4 benchmarks — Winogrande, MMLU, ARC AI2, GSM8K. The full table above shows each score with its evidence status.
0 of 4 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
DeepSeek Coder 33B has 4 tracked claims: 4 self-reported.