Qwen3 30B A3B 2507 by Qwen - a 30.5B MoE model with 3.3B active parameters, non-thinking mode, 131K context, and tool use.
Capabilities, design details, and architectural traits
Qwen3 30B A3B 2507 is a non-thinking instruction model in the Qwen3 line. It uses a mixture-of-experts design with a small active parameter set, and it is tuned for instruction following, multilingual understanding, and agentic tool use.
| Trait | Documented detail specific to Qwen3 30B A3B 2507 |
|---|---|
| Operating mode | Runs in non-thinking mode and does not generate <think> blocks |
| MoE activation | Uses 3.3B active parameters per inference from a 30.5B MoE model |
| Primary tuning | Optimized for instruction following, multilingual understanding, and tool use |
| Task profile | Documented gains in reasoning, coding, alignment, and open-ended subjective tasks |
The model is described as a chat-oriented release rather than a reasoning-trace model. Its documented design centers on direct instruction responses with strong support for multilingual and agent-style usage.
Compared with the broader Qwen3 family, this release is defined by non-thinking output, sparse MoE activation, and a long context window. Those traits make it easy to separate from thinking variants and from standard dense chat models.
Independent evaluations · Artificial Analysis
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Alibaba | Anthropic | Anthropic |
| Release Date | July 29, 2025 | July 24, 2026 | June 9, 2026 |
| Knowledge Cutoff | Jun 2025 | May 2026 | - |
| Context & Limits | |||
| Context Window | 131K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.20 Best Input Pricing | $5 | $10 |
| Output Pricing | $0.80 Best Output Pricing | $25 | $50 |
| Modalities | |||
| Inputs | text | textimage | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 8.9 | 63.1 Best Intelligence Index | 62.1 |
| Coding Index | - | 78.0 Best Coding Index | 76.5 |
| Agentic Index | - | 59.2 Best Agentic Index | 56.6 |
Graduate-level reasoning and expert Q&A evaluation.
Extremely difficult logical reasoning and knowledge.
Logical reasoning over long context windows.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
