DeepSeek-Coder 1.5B: pricing, context window and benchmarks
DeepSeek-Coder 1.5B is a coding model from DeepSeek. It offers a 16,384-token context window, costs $0.58 per million input tokens and $3.16 per million output tokens, and averages 51.0 across the index's benchmark suite.
Pricing
DeepSeek-Coder 1.5B costs $0.58 per million input tokens and $3.16 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.23 per million tokens. Cached input is listed at $0.07 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.58 | per 1M tokens |
| Output | $3.16 | per 1M tokens |
| Cached input | $0.07 | per 1M tokens |
| Blended 3:1 | $1.23 | estimate per 1M tokens |
| 1M in + 1M out | $3.74 | 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 | 58.5 | 546 | 9th | |
| GPQA Diamond | 45.3 | 563 | 6th | |
| SWE-bench Verified | 39.0 | 528 | 12th | |
| HumanEval | 53.3 | 552 | 8th | |
| MATH-500 | 48.8 | 565 | 6th | |
| AIME 2025 | 38.9 | 548 | 9th | |
| LiveCodeBench | 37.3 | 554 | 8th | |
| IFEval | 61.6 | 557 | 7th | |
| TruthfulQA | 57.5 | 550 | 8th | |
| HellaSwag | 69.2 | 544 | 9th | |
| MMMU | 51.9 | 532 | 11th | |
| Arena Elo | 1198.5 | 556 | 7th |
What DeepSeek-Coder 1.5B is good for
DeepSeek-Coder 1.5B 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 $0.58 per million input tokens, it sits in the mid-range of the index, and the 16,384-token context window is enough for large files and multi-file diffs.
Strengths
- Input price of $0.58 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
- Context window of 16,384 tokens is small for long-document work.
- 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 |
|---|---|---|---|---|
| MiniMax-Text 1.5B | MiniMax | 32768 | $0.58 | 51.5 |
| Phi-3.5 8B | Microsoft | 16384 | $0.59 | 51.2 |
| DeepSeek-Coder 1.5B | DeepSeek | 16384 | $0.58 | 51.0 |
| Nemotron 4 1.5B | NVIDIA | 32768 | $0.49 | 50.8 |
| Llama 3.1 1.5B | Meta AI | 4096 | $0.53 | 50.7 |
| Hunyuan-Large 1.5B | Tencent | 8192 | $0.54 | 50.5 |
FAQ
How much does DeepSeek-Coder 1.5B cost?
DeepSeek-Coder 1.5B is listed at $0.58 per million input tokens and $3.16 per million output tokens. Cached input is $0.073 per million tokens. A blended 3:1 input-to-output mix is about $1.23 per million tokens.
What is the context window of DeepSeek-Coder 1.5B?
DeepSeek-Coder 1.5B has a context window of 16,384 tokens and can emit up to 16,384 output tokens per request in this index.
Is DeepSeek-Coder 1.5B open weights?
Yes. DeepSeek-Coder 1.5B is listed as an open-weight model under a open weights licence.
How does DeepSeek-Coder 1.5B score on benchmarks?
DeepSeek-Coder 1.5B averages 51.0 across the scored benchmarks in this index, ranking #547 of 600 models. Its strongest relative result is Arena Elo at 1198.5.
What is DeepSeek-Coder 1.5B best for?
DeepSeek-Coder 1.5B 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 $0.58 per million input tokens, it sits in the mid-range of the index, and the 16,384-token context window is enough for large files and multi-file diffs.