Step 3.7 Flash vs Qwen 3 VL 235 B A22 B Instruct: cost comparison
Prices updated 2026-11-05 · computed from live provider data
Short answer: Step 3.7 Flash is cheaper on a
3:1 input/output blend — $0.438 per million tokens versus
$0.632, a 31% difference.
Both offer the same 262,144-token context window, so cost is the deciding factor.
Head to head
| Step 3.7 Flash | Qwen 3 VL 235 B A22 B Instruct | |
|---|---|---|
| Input, per 1M tokens | $0.200 | $0.210 |
| Output, per 1M tokens | $1.15 | $1.90 |
| Blended 3:1, per 1M | $0.438 | $0.632 |
| Context window | 262,144 | 262,144 |
| Vendor | empiriolabs | qwen |
Monthly cost at real volumes
| Workload (input / output) | Step 3.7 Flash | Qwen 3 VL 235 B A22 B Instruct |
|---|---|---|
| 1M in / 250K out | $0.487 | $0.685 |
| 10M in / 2.5M out | $4.88 | $6.85 |
| 100M in / 25M out | $48.75 | $68.50 |
Which should you pick?
Because the context windows are identical, the cheaper option is the better default. Choose Step 3.7 Flash unless you are already standardised on qwen tooling and want to avoid a second integration.