Codestral 2 vs DeepSeek-V3
A complete head-to-head: pricing, context window, benchmarks, modality coverage, and openness, with a programmatic verdict synthesized from the underlying data.
Updated
Codestral 2 specs · DeepSeek-V3 specs- PriceCodestral 2
Codestral 2 is roughly 1.2× cheaper on output tokens ($0.90 vs $1.10 per 1M).
- Context windowCodestral 2
Codestral 2 accepts 256K tokens vs 128K, 2.0× the room for long documents and codebases.
- BenchmarksCodestral 2
Codestral 2 leads in 1 of 1 shared benchmarks; the biggest gap is on HumanEval (Python coding), where it scores 92.0 vs 90.0.
- ModalitiesTie
Both handle text.
- OpennessTie
Both ship open weights, self-host either one.
On balance Codestral 2 edges ahead, winning 3 of 5 categories against DeepSeek-V3's 0. Codestral 2 is roughly 1.2× cheaper on output tokens ($0.90 vs $1.10 per 1M). Codestral 2 accepts 256K tokens vs 128K, 2.0× the room for long documents and codebases.
Codestral 2 leads in 1 of 1 shared benchmarks; the biggest gap is on HumanEval (Python coding), where it scores 92.0 vs 90.0. Both target the same set of modalities (text), so the deciding factors are price, context, and raw quality. Both ship open weights, self-host either one.
Both shipped within roughly a month of each other in 2025, so they share the same generation of training data and tooling. Codestral 2 is usually picked for code completion and coding llm workloads, while DeepSeek-V3 sees more deployments in open source llm and commercial use llm. If pricing matters more than every last benchmark point, run the numbers in the calculator below before committing.
Side-by-side specs
| Spec | Codestral 2 | DeepSeek-V3 |
|---|---|---|
| Provider | Mistral | DeepSeek |
| Released | Jan 2025 | Dec 2024 |
| Modalities | text | text |
| Context window | 256K tokens | 128K tokens |
| Max output | , | , |
| Input · 1M | $0.30 / 1M tokens | $0.27 / 1M tokens |
| Output · 1M | $0.90 / 1M tokens | $1.10 / 1M tokens |
| Knowledge cutoff | , | 2024-07 |
| Open weights | Yes (Mistral Non-Production License) | Yes (MIT) |
| API available | Yes | Yes |
Pricing at scale
What you'd actually pay at typical workloads. Numbers come from each model's published per-million-token rates.
- Light usage, 100k in / 50k out per day$2.25 vs $2.46
- Heavy usage, 1M in / 500k out per day$22.50 vs $24.60
- RAG workload, 5M in / 200k out per day$50.40 vs $47.10
Light usage, 100k in / 50k out per day: $2.25 vs $2.46 per month, model A comes out ahead. Heavy usage, 1M in / 500k out per day: $22.50 vs $24.60 per month, model A comes out ahead. RAG workload, 5M in / 200k out per day: $50.40 vs $47.10 per month, model B comes out ahead.
Estimated spend for the listed models at your usage. Numbers are derived from each model's published per-million-token rates.
- Codestral 2$0.075
- DeepSeek-V3$0.082
Benchmarks compared
Only sourced numbers. Where a benchmark is missing for one model we show the available value rather than fabricating the other.
- MMLU,88.5
- HumanEval92.090.0
Codestral 2 fits when…
- Fast inline completion
- FIM support
- Self-hostable for IP-sensitive teams
- Cost-sensitive workloads, 1.2× cheaper than DeepSeek-V3 on output tokens.
- Long-context tasks, handles 256K tokens vs 128K for DeepSeek-V3.
DeepSeek-V3 fits when…
- Frontier-level quality at open-weight prices
- MIT license, clean commercial use
- Cheap to serve via MoE architecture
Consider Mistral Large 2
Mistral's flagship API model, strong on code and reasoning, EU-friendly hosting.
Frequently asked
Is Codestral 2 or DeepSeek-V3 cheaper?
Codestral 2 is cheaper at $0.90 / 1M tokens per million output tokens, vs $1.10 / 1M tokens for DeepSeek-V3.Which has the larger context window?
Codestral 2 accepts 256K tokens vs 128K for DeepSeek-V3.Is Codestral 2 or DeepSeek-V3 better for coding?
Both Codestral 2 and DeepSeek-V3 are competitive on coding benchmarks. See each model's individual spec page for HumanEval and SWE-bench scores where published. For an opinionated pick, consult our Best LLM for Coding ranking.Are either of these models open source?
Both ship with open weights. Codestral 2 is licensed under Mistral Non-Production License; DeepSeek-V3 under MIT.When were Codestral 2 and DeepSeek-V3 released?
Codestral 2 was released by Mistral on 2025-01-14. DeepSeek-V3 was released by DeepSeek on 2024-12-26.
The week's AI launches, in your inbox.
One short email every Friday, new models, leaks, and quietly-shipped APIs you missed.