Qwen3-Coder Base: pricing, context window and benchmarks
Qwen3-Coder Base is a coding model from Alibaba Cloud. It offers a 200,000-token context window, costs $3.99 per million input tokens and $18.04 per million output tokens, and averages 76.6 across the index's benchmark suite.
Pricing
Qwen3-Coder Base costs $3.99 per million input tokens and $18.04 per million output tokens. A blended 3:1 input-to-output mix works out to about $7.50 per million tokens. Cached input is listed at $0.35 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $3.99 | per 1M tokens |
| Output | $18.04 | per 1M tokens |
| Cached input | $0.35 | per 1M tokens |
| Blended 3:1 | $7.50 | estimate per 1M tokens |
| 1M in + 1M out | $22.03 | 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 | 78.3 | 235 | 61th | |
| GPQA Diamond | 72.2 | 216 | 64th | |
| SWE-bench Verified | 64.0 | 232 | 61th | |
| HumanEval | 78.5 | 282 | 53th | |
| MATH-500 | 82.4 | 156 | 74th | |
| AIME 2025 | 73.7 | 193 | 68th | |
| LiveCodeBench | 68.0 | 227 | 62th | |
| IFEval | 85.7 | 168 | 72th | |
| TruthfulQA | 80.8 | 162 | 73th | |
| HellaSwag | 85.5 | 228 | 62th | |
| MMMU | 73.4 | 208 | 65th | |
| Arena Elo | 1409.2 | 185 | 69th |
What Qwen3-Coder Base is good for
Qwen3-Coder Base 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 $3.99 per million input tokens, it sits in the mid-range of the index, and the 200,000-token context window is enough for large files and multi-file diffs.
Strengths
- Large 200,000-token context window for long documents and agent traces.
- Input price of $3.99 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
- Accepts more than text, including image input where the provider supports it.
- Coding benchmarks are a relative strength within its tier.
Trade-offs
- No standout weakness in the indexed specifications, but real-world behaviour still needs testing.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Llama 3.3 Instruct | Meta AI | 128000 | $4.33 | 76.6 |
| Nemotron 4 | NVIDIA | 65536 | $4.18 | 76.6 |
| Qwen3-Coder Base | Alibaba Cloud | 200000 | $3.99 | 76.6 |
| Claude Sonnet 4.5 Mini | Anthropic | 32768 | $11.43 | 76.5 |
| Qwen3-Max Mini | Alibaba Cloud | 128000 | $11.13 | 76.5 |
| Yi-Lightning Instruct | 01.AI | 128000 | $4.16 | 76.5 |
FAQ
How much does Qwen3-Coder Base cost?
Qwen3-Coder Base is listed at $3.99 per million input tokens and $18.04 per million output tokens. Cached input is $0.35 per million tokens. A blended 3:1 input-to-output mix is about $7.50 per million tokens.
What is the context window of Qwen3-Coder Base?
Qwen3-Coder Base has a context window of 200,000 tokens and can emit up to 8,192 output tokens per request in this index.
Is Qwen3-Coder Base open weights?
Yes. Qwen3-Coder Base is listed as an open-weight model under a open weights licence.
How does Qwen3-Coder Base score on benchmarks?
Qwen3-Coder Base averages 76.6 across the scored benchmarks in this index, ranking #205 of 600 models. Its strongest relative result is Arena Elo at 1409.2.
What is Qwen3-Coder Base best for?
Qwen3-Coder Base 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 $3.99 per million input tokens, it sits in the mid-range of the index, and the 200,000-token context window is enough for large files and multi-file diffs.