Qwen3 Coder 30B A3B Instruct
Qwen · released Jul 31, 2025
Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...
Specification
- Context window
- 262K
- Max output
- 33K
- Knowledge cutoff
- Jun 30, 2025
- Parameters
- 30.5B
- Licence
- apache-2.0
- Serving providers
- 5
- Moderated
- No
- Uptime
- 100.0%
Intelligence
13.6
45th percentile
Coding
—
Coding Index
Agentic
—
Agentic Index
Output speed
76 t/s
Median across providers
Latency
876ms
Time to first token
Cost per task
$0.01
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- MMLU-Pro70.6%
- GPQA Diamond51.6%
- LiveCodeBench40.3%
- τ²-bench (Telecom)34.5%
- IFBench32.7%
- AA-LCR (long context)31.7%
- AIME 202529.0%
- SciCode27.8%
- Terminal-Bench Hard15.2%
- Humanity's Last Exam3.8%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| MMLU-Pro | 70.6% |
| GPQA Diamond | 51.6% |
| LiveCodeBench | 40.3% |
| τ²-bench (Telecom) | 34.5% |
| IFBench | 32.7% |
| AA-LCR (long context) | 31.7% |
| AIME 2025 | 29.0% |
| SciCode | 27.8% |
| Terminal-Bench Hard | 15.2% |
| Humanity's Last Exam | 3.8% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Mistral Medium 3.1Mistral AI14.7
- Solar Pro 3Upstage14.5
- Llama 4 MaverickMeta14.5
- 14.4
- Ling-2.6-flashInclusionAI14.2
- Gemini 2.5 FlashGoogle14.2
- 13.8
- Qwen3 Coder 30B A3B InstructQwen13.6
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Mistral Medium 3.1 | 14.7 |
| Solar Pro 3 | 14.5 |
| Llama 4 Maverick | 14.5 |
| Qwen3 VL 235B A22B Instruct | 14.4 |
| Ling-2.6-flash | 14.2 |
| Gemini 2.5 Flash | 14.2 |
| Qwen3 Next 80B A3B Instruct | 13.8 |
| Qwen3 Coder 30B A3B Instruct | 13.6 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $0.07
- Output / 1M tokens
- $0.27
- Cached input / 1M
- Not offered
- Blended 3:1
- $0.12
One task, estimated
$0.01
- Input tokens
- 50,000
- Output tokens
- 25,000
- Profile
- Standard
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 1,111
- Win rate
- 54.7%
- Strongest at
- Data visualisation
- Tournaments
- 64
Elo by category
Dot position on a 1,075–1,125 scale · Elo has no meaningful zero
- 1,111
- 1,110
- 1,081
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| Data visualisation | 1111 | 54.7% |
| Websites | 1110 | 57.1% |
| UI components | 1081 | 54.1% |