DeepSeek-Coder 7B: pricing, context window and benchmarks
DeepSeek-Coder 7B is a coding model from DeepSeek. It offers a 32,768-token context window, costs $1.14 per million input tokens and $4.37 per million output tokens, and averages 57.3 across the index's benchmark suite.
Pricing
DeepSeek-Coder 7B costs $1.14 per million input tokens and $4.37 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.95 per million tokens. Cached input is listed at $0.16 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $1.14 | per 1M tokens |
| Output | $4.37 | per 1M tokens |
| Cached input | $0.16 | per 1M tokens |
| Blended 3:1 | $1.95 | estimate per 1M tokens |
| 1M in + 1M out | $5.51 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 61.7 | 506 | 16th | |
| GPQA Diamond | 54.1 | 478 | 20th | |
| SWE-bench Verified | 43.6 | 483 | 20th | |
| HumanEval | 62.0 | 468 | 22th | |
| MATH-500 | 56.9 | 498 | 17th | |
| AIME 2025 | 43.8 | 511 | 15th | |
| LiveCodeBench | 48.3 | 462 | 23th | |
| IFEval | 70.1 | 459 | 24th | |
| TruthfulQA | 63.8 | 471 | 22th | |
| HellaSwag | 73.2 | 474 | 21th | |
| MMMU | 53.2 | 517 | 14th | |
| Arena Elo | 1260.7 | 473 | 21th |
What DeepSeek-Coder 7B is good for
DeepSeek-Coder 7B is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $1.14 per million input tokens, it sits in the mid-range of the index, and the 32,768-token context window is enough for large files and multi-file diffs.
Strengths
- Input price of $1.14 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
- Coding benchmarks are a relative strength within its tier.
Trade-offs
- Average benchmark score is below the index median; test it on your own task.
- Less reliable on open-ended reasoning and complex multi-step instructions.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Llama 3 DeepInfra 8B | DeepInfra | 8192 | $1.05 | 57.4 |
| Llama 3.1 7B | Meta AI | 8192 | $1.01 | 57.4 |
| DeepSeek-Coder 7B | DeepSeek | 32768 | $1.14 | 57.3 |
| Llama 3 Anyscale 8B | Anyscale | 32768 | $0.97 | 57.3 |
| Llama 3 Lambda 8B | Lambda | 4096 | $1.03 | 57.2 |
| Mistral Small 3 8B | Mistral AI | 8192 | $0.99 | 57.2 |
FAQ
How much does DeepSeek-Coder 7B cost?
DeepSeek-Coder 7B is listed at $1.14 per million input tokens and $4.37 per million output tokens. Cached input is $0.16 per million tokens. A blended 3:1 input-to-output mix is about $1.95 per million tokens.
What is the context window of DeepSeek-Coder 7B?
DeepSeek-Coder 7B has a context window of 32,768 tokens and can emit up to 32,768 output tokens per request in this index.
Is DeepSeek-Coder 7B open weights?
Yes. DeepSeek-Coder 7B is listed as an open-weight model under a open weights licence.
How does DeepSeek-Coder 7B score on benchmarks?
DeepSeek-Coder 7B averages 57.3 across the scored benchmarks in this index, ranking #481 of 600 models. Its strongest relative result is Arena Elo at 1260.7.
What is DeepSeek-Coder 7B best for?
DeepSeek-Coder 7B is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $1.14 per million input tokens, it sits in the mid-range of the index, and the 32,768-token context window is enough for large files and multi-file diffs.