Step 3.5 Flash is 8.5× cheaper per million tokens (blended). Qwen3.8 27B has the larger context window (1,000,000 vs 262,144 tokens, 3.8×).
Qwen3.8 27B vs Step 3.5 Flash
Live catalog fields. Best-in-row is highlighted.
Qwen3.8 27Bopen
Qwen · qwen__qwen3.8-27b
Step 3.5 Flashopen
Stepfun · stepfun__step-3.5-flash
Identity
| Field | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
|---|---|---|
| Provider | Qwen | Stepfun |
| Slug | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
| Status | live | live |
| Open weights | Yes | Yes |
| License | — | — |
| HuggingFace | Qwen/Qwen3.8-27B | stepfun-ai/Step-3.5-Flash |
Quality
| Field | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
|---|---|---|
| Quality index (0–100) | 87.4 | — |
Cost
| Field | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
|---|---|---|
| Input $ / 1M tokens | $0.42 | $0.10 |
| Cached input $ / 1M | $0.09 | — |
| Output $ / 1M tokens | $3.00 | $0.30 |
Context
| Field | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
|---|---|---|
| Max context tokens | 1,000,000 | 262,144 |
| Max output tokens | 131,072 | 65,536 |
Modalities
| Field | qwen__qwen3.8-27b | stepfun__step-3.5-flash |
|---|---|---|
| Input | text, image, video | text |
| Output | text | text |