Qwen3 235B A22B Instruct 2507 API Pricing
Retired October 10, 2026: Alibaba took qwen3-235b-a22b-instruct-2507 offline on that date, and calls to it now fail; the model is gone from Alibaba's pricing page. Alibaba recommends Qwen3.7 Plus instead. Qwen3 235B A22B Instruct 2507 is an Alibaba Qwen3 text model priced from the Singapore/International row of Model Studio. The verified rate is $0.23/M input and $0.92/M output. The 2507 Instruct checkpoint is the non-thinking 235B A22B update.
Qwen3 235B A22B Instruct 2507 API cost calculator.
Live calculator pre-loaded with the verified International Qwen3 235B A22B Instruct 2507 rates. Use it for invoice checks, agent traces, and scenario planning before large Model Studio runs.
Qwen3 235B A22B Instruct 2507 cost on real workloads.
Support answer
Policy analysis
Function rewrite
Document summary
Counts use a chars-per-token calibration measured on the vendor's own published tokenizer (Qwen/Qwen3.5-397B-A17B, 2026-06-10). English prose is typically within a few percent; code and non-Latin scripts tokenize heavier. For billing-exact counts use the vendor's count-tokens API.
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Qwen3 235B A22B Instruct 2507 Current | $0.23 | $0.92 | $0.285 agentic 92/8 | 262K | High-end non-thinking Qwen calls |
| Qwen3 Next 80B A3B Thinking | $0.15 | $1.20 | $0.234 retired | 262K | Qwen3 sibling |
| Qwen3 Next 80B A3B Instruct | $0.15 | $1.20 | $0.234 retired | 262K | Qwen3 sibling |
| Qwen3 235B A22B Thinking 2507 | $0.23 | $2.30 | $0.396 retired | 262K | Qwen3 sibling |
| Qwen3 30B A3B Thinking 2507 | $0.20 | $2.40 | $0.376 retired | 262K | Qwen3 sibling |
| Qwen3 30B A3B Instruct 2507 | $0.20 | $0.80 | $0.248 retired | 262K | Qwen3 sibling |
| Qwen3 30B A3B | $0.20 | $0.80 | $0.248 retired | 128K | Qwen3 sibling |
| Qwen3 8B | $0.18 | $0.70 | $0.222 retired | 128K | Qwen3 sibling |
| Qwen3.7 Plus | $0.40 | $1.60 | $0.496 +74% · promo | 1M | Current Plus tier |
| Qwen3.7 Max | $2.50 | $7.50 | $2.90 +10× | 1M | Current Max flagship |
Audit links
Qwen3 235B A22B Instruct 2507 has published scores on 2 benchmarks we track; its best placing is #86 of 105 on ARC-AGI-2. All of these scores come from independent leaderboards, not from the lab.
| Benchmark | Score | Rank | Leader | Source |
|---|---|---|---|---|
| ARC-AGI-2 Reasoning | 1.3% ARC Prize verified, best result | #86 of 105 | GPT-6 Astra · 95% | Official leaderboard ARC Prize Foundation |
| ARC-AGI-1 Reasoning | 11% ARC Prize verified, best result | #89 of 99 | 6 tied · 98.5% | Official leaderboard ARC Prize Foundation |
Qwen3 235B A22B Instruct 2507 API pricing FAQ.
Short answers for teams comparing Qwen3 235B A22B Instruct 2507 against other current Qwen3 text models.
Q · 01 When did Qwen3 235B A22B Instruct 2507 stop working? +
qwen3-235b-a22b-instruct-2507 would go offline on October 10, 2026 (00:00, UTC+8) in the Beijing and Singapore regions, and it has: on October 21, 2026 the model is no longer on Alibaba's pricing page, and calls to it fail. Its notice recommends Qwen3.7 Plus. The rates on this page are the last Alibaba published.