Codestral 25.08 vs Qwen3 VL 235B A22B Instruct
Price and specs side by side · USD per 1M tokens · verified 2026-11-22
// side by side
| metric | Codestral 25.08 | Qwen3 VL 235B A22B Instruct |
|---|
| input / 1M | $0.30 | $0.20 |
| output / 1M | $0.90 | $0.88 |
| blended (3:1) | $0.45 | $0.37 |
| cache read / 1M | $0.03 | $0.11 |
| context | 256K | 262.1K |
| max output | 8.2K | 32.8K |
| input types | text · pdf | text · image |
| reasoning | no | no |
| tool calling | yes | yes |
| structured output | yes | yes |
| open weights | no | yes |
| released | 2025-07-30 | 2025-09-23 |
| knowledge cutoff | n/a | 2025-03-31 |
| lab | Mistral | Alibaba (Qwen) |
// monthly cost scenarios
| workload | tokens in / out | Codestral 25.08 | Qwen3 VL 235B A22B Instruct |
|---|
| Chatbot | 10M / 2M | $4.80 | $3.76 |
| RAG app | 50M / 5M | $19.50 | $14.40 |
| Coding agent | 20M / 10M | $15.00 | $12.80 |
// verdict
- Qwen3 VL 235B A22B Instruct is 17.8% cheaper than Codestral 25.08 on a blended 3:1 input/output basis.
- Qwen3 VL 235B A22B Instruct has the larger context window (262.1K vs 256K tokens).
- Only Qwen3 VL 235B A22B Instruct accepts image input.
- Chatbot (10M input / 2M output tokens per month): Qwen3 VL 235B A22B Instruct costs $3.76 vs $4.80 for Codestral 25.08, saving $1.04 per month.
- RAG app (50M input / 5M output tokens per month): Qwen3 VL 235B A22B Instruct costs $14.40 vs $19.50 for Codestral 25.08, saving $5.10 per month.
- Coding agent (20M input / 10M output tokens per month): Qwen3 VL 235B A22B Instruct costs $12.80 vs $15.00 for Codestral 25.08, saving $2.20 per month.
// Codestral 25.08 or Qwen3 VL 235B A22B Instruct: which one to choose
- Choose Qwen3 VL 235B A22B Instruct for cost-sensitive, high-volume workloads.
- Choose Qwen3 VL 235B A22B Instruct for long documents and large codebases (262.1K-token context).
- Choose Qwen3 VL 235B A22B Instruct if you need image input.
- Choose Qwen3 VL 235B A22B Instruct if you want open weights you can self-host.