AlibabaQwen3-VL 30B-A3B
Best configuration ranks #407 of 610 on the BenchLeader Index at 44.8 ±3.6 (thinking reasoning effort). Released 5 Oct 2025.
- Blended price
- $0.750/M
- $0.200 in · $2.40 out
- Output speed
- 111 tok/s
- First answer
- 20 s
- first token 2.15 s
- Context
- 262k
- Overall index45
- Agents & tools38
- Knowledge37
- Instruction following47
- Multimodal46
- Composite41
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| thinkingbest | 44.8 | #407 | 111 tok/s | 20 s | $0.0008 | Agents & tools 38 · Composite 41 · Instruction following 47 · Knowledge 37 · Multimodal 46 |
| default | 43.1 | #445 | 111 tok/s | 2.15 s | $0.0002 | Agents & tools 39 · Composite 39 · Instruction following 37 · Knowledge 34 · Multimodal 46 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | thinking | default | Source |
|---|---|---|---|
| GPQA Diamond (AA)not in index | 72.0%#271 | 69.5%#291 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 8.9%#272 | 6.3%#324 | Artificial Analysis |
Agents & tools
| Benchmark | thinking | default | Source |
|---|---|---|---|
| Terminal-Bench Hard | 5.3%#263 | 6.1%#257 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 19.9%#321 | 19.0%#329 | Artificial Analysis |
Knowledge
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA-Omniscience | -57.6#370 | -63.4#412 | Artificial Analysis |
Instruction following
| Benchmark | thinking | default | Source |
|---|---|---|---|
| IFBench | 45.1%#204 | 33.1%#317 | Artificial Analysis |
Multimodal
| Benchmark | thinking | default | Source |
|---|---|---|---|
| MMMU-Pro | 61.9%#170 | 62.1%#164 | Artificial Analysis |
Composite
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA Intelligence Index | 9.4#328 | 7.9#371 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| SiliconFlow | 14 tok/s | 2.16 s | $0.290 | $1.00 | 262k | fp8 |
| Alibaba Cloud Int. | 8 tok/s | 0.48 s | $0.130 | $0.520 | 131k | – |
| NovitaAI | 8 tok/s | 0.74 s | $0.200 | $0.700 | 131k | bf16 |
| DeepInfra | 8 tok/s | 0.63 s | $0.150 | $0.600 | 262k | fp8 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0008 | – | 22.9 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0038 | – | 25.6 s |
| Code edit | 6,000 / 1,500 | $0.0048 | – | 33.8 s |
| Agentic coding session | 60,000 / 4,000 | $0.022 | – | 56.3 s |
| Structured extraction | 2,000 / 200 | $0.0009 | – | 22.0 s |
See also
Data as of 9 Sept 2026. Compare these configurations.