DeepSeek V3
ActiveOpen-weight frontier model competitive with GPT-4o and Claude Sonnet at fraction of training cost.
Overview
DeepSeek V3 is a 671B mixture-of-experts model (37B active parameters) trained for a reported $6M in compute — dramatically less than comparable frontier models. It matches GPT-4o and Claude Sonnet on most standard benchmarks.
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| Aider PolyglotCoding | 55.1% pass@2 | Third-party Papers With Code |
| GSM8KMath | 97.1% accuracy | Self-reported DeepSeek tech report |
| HumanEvalCoding | 90.2pass@1 % | Self-reported DeepSeek tech report |
| MMLUGeneral knowledge | 88.5% accuracy | Self-reported DeepSeek tech report |
| MMLU-ProGeneral knowledge | 75.9% accuracy | Self-reported DeepSeek tech report |
Integrations & tooling support
- Tool calling
- Supported
- Structured outputs
- Supported
Price vs quality
Solid value
Competent capability at a low price.
- Quality percentile
- 65%
- Effective price
- $0.8925/1M
- Pricing breakdown
- $0.27/1M in
$1.1/1M out
vs 5 benchmarks
/ 1M tokens (input + 3× output)
Community ratings
No ratings yet. Be the first to rate DeepSeek V3.
Rate DeepSeek V3
Sign in to rate and review.
Comments
Sign in to leave a comment.