Model pricing
Qwen3 235B Thinking pricing
Qwen3 235B Thinking by Alibaba costs
Qwen3 235B Thinking pricing
.5 per 1M input tokens and $6 per 1M output tokens (cached input not recorded, batch input not recorded) as of 2026-06-13. A 1,000-input/500-output request costs $0.0045. The lowest-cost alternative with at least this context window and the same recorded reasoning category is GPT-6 Luna Pro at $0.1/$0.5; capability equivalence is not guaranteed.Notes
Qwen3 thinking mode
Key pricing facts
| Fact | Value |
|---|---|
| Input / 1M tokens | $1.5 |
| Output / 1M tokens | $6 |
| Cached input / 1M | Not recorded |
| Batch input / output / 1M | Not recorded |
| Context window | 256,000 |
| Provider | Alibaba |
| Last verified | 2026-06-13 |
| Snapshot updated | 2026-06-13 |
Last updated: · Source: No official provider pricing URL recorded · Methodology
Frequently asked questions
What does Qwen3 235B Thinking cost per million tokens?
Input costs $1.5 and output costs $6 per million tokens. The 1,000-input/500-output example costs $0.0045, before discounts, taxes, hidden reasoning tokens or context-tier surcharges.
Does Qwen3 235B Thinking have cached or batch pricing?
Cached input: not separately recorded. Batch: not recorded in this dataset. Eligibility and cache-write charges depend on the provider; these discounts are not automatically stacked.
What is a cheaper alternative to Qwen3 235B Thinking?
GPT-6 Luna Pro costs $0.1 input and $0.5 output per million tokens. This comparison matches context capacity and reasoning category, not benchmark quality. Test your workload before replacing a model.
How often are these prices verified?
The live pricing table synchronizes stored model records daily; public file snapshots update when the site is rebuilt. Synchronization is not independent provider verification. Snapshot updated 2026-06-13. The recorded model verification date is 2026-06-13.