Qwen: Qwen3 30B A3B Thinking 2507
Qwen3 30B A3B Thinking 2507 is a compact thinking-mode model with tool use and a modest 32K context window, but the absence of benchmark data limits our ability to assess its real-world capability — it scores conservatively despite Qwen's generally strong model family.
Assessment date: March 14, 2026
Our methodology takes into account a range of factors including pricing, functionality, capabilities, benchmark performance, and real-world applicability. Rankings are reviewed and updated regularly as new models are released. Issues with our rankings? Contact us
Qwen3-30B-A3B-Thinking-2507 is a 30B parameter Mixture-of-Experts reasoning model optimized for complex tasks requiring extended multi-step thinking. The model is designed specifically for “thinking mode,” where internal reasoning traces are separated from final answers. Compared to earlier Qwen3-30B releases, this version improves performance across logical reasoning, mathematics, science, coding, and multilingual benchmarks. It also demonstrates stronger instruction following, tool use, and alignment with human preferences. With higher reasoning efficiency and extended output budgets, it is best suited for advanced research, competitive problem solving, and agentic applications requiring structured long-context reasoning.
Capabilities
Architecture
| Modality | Text → Text |
| Tokenizer | Qwen3 |
| Parameters | 30B |
Model Information
Pricing
| Token Type | Cost per 1M tokens | Cost per 1K tokens |
|---|---|---|
| Input | $0.05 | $0.000051 |
| Output | $0.34 | $0.000340 |
Live Performance
Live endpoint metrics — refreshed every 30 minutes.
