Qwen3 30B A3B Thinking 2507
Qwen released 2025-08-28
DeepSeek V4 Flash 0731 is both better and cheaper.
It scores +37.2 higher and costs 86% less ($0.105/M against $0.750/M) on this workload — and it does everything this model does.
KAT-Coder-Pro V2 is cheaper still (30% less) but drops no extended reasoning.
Every price dimension
| Input | $0.200/M |
|---|---|
| Output | $2.40/M |
Independent scores
| Intelligence | 14.6 |
|---|---|
| Coding | 12.1 |
| Agentic | 1.8 |
Sources are listed separately rather than averaged. Across the 62 models both have scored they correlate at r = 0.806 — close agreement overall, but four models rank very differently between them, and a blended score would hide exactly those.
Capability
| Context window | 82K |
|---|---|
| Max output | 33K |
| Input modes | text |
| Tool use | yes |
| Reasoning | always on |
| Knowledge cutoff | 2025-06-30 |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-08-24 |
| Quality data | verified |
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...