Qwen 3 235 B A22 B Thinking 2507 Fp 8 vs Mistral Small 3.1 24 B Instruct: cost comparison
Prices updated 2026-11-05 · computed from live provider data
Short answer: Mistral Small 3.1 24 B Instruct is cheaper on a
3:1 input/output blend — $0.402 per million tokens versus
$1.31, a 69% difference.
Qwen 3 235 B A22 B Thinking 2507 Fp 8 offers the larger context window at 262,141 tokens, against 128,000 for Mistral Small 3.1 24 B Instruct.
Head to head
| Qwen 3 235 B A22 B Thinking 2507 Fp 8 | Mistral Small 3.1 24 B Instruct | |
|---|---|---|
| Input, per 1M tokens | $0.350 | $0.351 |
| Output, per 1M tokens | $4.20 | $0.555 |
| Blended 3:1, per 1M | $1.31 | $0.402 |
| Context window | 262,141 | 128,000 |
| Vendor | gmi | mistralai |
Monthly cost at real volumes
| Workload (input / output) | Qwen 3 235 B A22 B Thinking 2507 Fp 8 | Mistral Small 3.1 24 B Instruct |
|---|---|---|
| 1M in / 250K out | $1.40 | $0.490 |
| 10M in / 2.5M out | $14.00 | $4.90 |
| 100M in / 25M out | $140 | $48.98 |
Which should you pick?
Pick Qwen 3 235 B A22 B Thinking 2507 Fp 8 when you need the longer context window (262,141 vs 128,000 tokens). Pick Mistral Small 3.1 24 B Instruct when price per token dominates your workload — high-volume classification, extraction, or batch summarisation.