Leaderboard · Model · Updated
Qwen2.5-Coder 32B
Coder-tuned Qwen2.5 — among the strongest open coding models at <50B.
At a glance
| Provider | Alibaba |
|---|---|
| Released | 2024-11 |
| Tier | open-weights |
| License | Open-weights (Apache-2.0) |
| Modalities | text |
| Context window | 131.072k tokens |
| Max output | 8.192k tokens |
| API price · input | $0.18 / 1M tokens |
| API price · output | $0.18 / 1M tokens |
| Hugging Face | Qwen/Qwen2.5-Coder-32B-Instruct |
Benchmark performance
How Qwen2.5-Coder 32B stacks up against the current leader (GPT-5) and the median model in the leaderboard:
| Benchmark | Qwen2.5-Coder 32B | GPT-5 | Median |
|---|---|---|---|
| Chatbot Arena Elo | N/A | 1410 | 1320 |
| MMLU-Pro | 68.4 | 86.8 | 78.0 |
| GPQA Diamond | 40.0 | 87.3 | 65.0 |
| MATH | 83.1 | 96.7 | 78.3 |
| HumanEval | 92.7 | 95.1 | 92.0 |
| SWE-Bench Verified | N/A | 74.9 | 49.0 |
Numbers compiled from provider technical reports and Chatbot Arena. See methodology for the composite-score formula.
OpenRouter routes your requests across Qwen2.5-Coder 32B, GPT-5, Claude, Gemini, and 100+ other models behind a single API key — pay-as-you-go, no monthly minimum. Try OpenRouter → (affiliate · supports this site)
What does Qwen2.5-Coder 32B cost in practice?
API pricing is $0.18 per 1M input tokens and $0.18 per 1M output tokens. Assuming a 50/50 input/output split, here is what that looks like at three workload sizes:
| Volume | Per day | Per month | Per year |
|---|---|---|---|
| 1M tokens/day | $0.18 | $5.40 | $65.70 |
| 10M tokens/day | $1.80 | $54.00 | $657.00 |
| 100M tokens/day | $18.00 | $540.00 | $6,570 |
Strengths & weaknesses
Where it shines
- HumanEval: 92.7 (rank #6 of 30, top tier)
Where it lags
- GPQA: 40.0 (rank #29 of 29, below average)
- MMLU-Pro: 68.4 (rank #25 of 29, below average)
Best alternatives
The closest models to Qwen2.5-Coder 32B by tier and benchmark score:
| Model | Score | $ in / out | Context | Action |
|---|---|---|---|---|
| DeepSeek V3 DeepSeek |
68.0 | $0.27 / $1.10 | 128k | Try → · vs Qwen2.5-Coder |
| Phi-4 Microsoft |
71.2 | $0.07 / $0.14 | 16.384k | Try → · vs Qwen2.5-Coder |
| Llama 3.1 405B Instruct Meta |
65.7 | $2.70 / $2.70 | 128k | Try → · vs Qwen2.5-Coder |
| Qwen2.5 72B Instruct Alibaba |
65.6 | $0.35 / $0.40 | 131.072k | Try → · vs Qwen2.5-Coder |
| Llama 3.3 70B Instruct Meta |
64.7 | $0.23 / $0.40 | 128k | Try → · vs Qwen2.5-Coder |
Frequently asked questions
Is Qwen2.5-Coder 32B a good model?
Qwen2.5-Coder 32B scores 68.8 on the llmrank.top composite (rank #17 of 30). It trails the frontier — for top performance, look at GPT-5 (86.0). Whether it's the right fit depends on your workload — see the use-case discussion on this page.
How much does Qwen2.5-Coder 32B cost?
Qwen2.5-Coder 32B is priced at $0.18 per 1M input tokens and $0.18 per 1M output tokens (USD). For a 10M-token-per-day workload split 50/50, that works out to roughly $1.80/day or $657.00/year.
What is Qwen2.5-Coder 32B's context window?
131.072k tokens. For very long inputs, consider a 1M+ context model like Gemini 2.5 Pro or GPT-4.1.
Is Qwen2.5-Coder 32B open source?
Yes — released under the Apache-2.0 license, weights are downloadable from the provider or Hugging Face.
What is Qwen2.5-Coder 32B's SWE-Bench score?
Qwen2.5-Coder 32B scores not reported on SWE-Bench Verified — the benchmark that measures real-world GitHub issue resolution. Its Chatbot Arena Elo is not reported.
What are the best alternatives to Qwen2.5-Coder 32B?
Closest alternatives by tier and score: DeepSeek V3, Phi-4, Llama 3.1 405B Instruct. See the alternatives section on this page for side-by-side numbers.
Related: DeepSeek V3 · Phi-4 · Llama 3.1 405B Instruct · Full leaderboard
Spotted out-of-date numbers? Open an issue — corrections usually ship within 24h.
Try Qwen2.5-Coder 32B now
Direct link to the official playground — or use OpenRouter to A/B test it against any other model on a single API key.