Qwen3 235B A22B Instruct 2507 API Prices Rise: Input Cost Up 66%

What Changed

Qwen has pushed through a notable price increase on Qwen3 235B A22B Instruct 2507. Input pricing moves from $0.09 to $0.149 per 1M tokens — a +66% jump. Output pricing also ticks up, from $0.55 to $0.598 per 1M tokens, a smaller but still real increase.

See the full pricing timeline on the Qwen3 235B A22B Instruct 2507 — live specs & price history.

Does This Matter for Your Workload?

The output increase is modest — most inference-heavy pipelines won't feel it much. The input hike is the story here. At 66% up, any workload that's input-heavy (long-context retrieval, document processing, large system prompts) will see a meaningful cost bump.

Quick sanity check: if you're currently pushing 100M input tokens per month, your monthly input bill moves from $9.00 to $14.90. At 1B tokens, that's $90 vs. $149. Not catastrophic for a 235B MoE model, but worth re-running your cost projections before your next billing cycle.

What to Do Now

• Audit your input-to-output ratio. If inputs dominate your usage, the impact is real.
• Compare alternatives. Our model comparison tools let you filter by price tier — useful if you're now on the wrong side of your budget.
• Check if output quality justifies the new rate. Qwen3 235B A22B remains a competitive large MoE model; the question is whether your specific tasks need this scale or if a smaller, cheaper model covers it.

Pricing changes like this are a good trigger to revisit model selection rather than defaulting to the same endpoint.

Ezra, Scout AI Team