DeepSeek V4.1 Flash 1017 vs Step 3.7 Flash: cost comparison
Prices updated 2026-11-07 · computed from live provider data
Short answer: DeepSeek V4.1 Flash 1017 is cheaper on a
3:1 input/output blend — $0.350 per million tokens versus
$0.438, a 20% difference.
DeepSeek V4.1 Flash 1017 offers the larger context window at 1,000,000 tokens, against 262,144 for Step 3.7 Flash.
Head to head
| DeepSeek V4.1 Flash 1017 | Step 3.7 Flash | |
|---|---|---|
| Input, per 1M tokens | $0.200 | $0.200 |
| Output, per 1M tokens | $0.800 | $1.15 |
| Blended 3:1, per 1M | $0.350 | $0.438 |
| Context window | 1,000,000 | 262,144 |
| Vendor | empiriolabs | empiriolabs |
Monthly cost at real volumes
| Workload (input / output) | DeepSeek V4.1 Flash 1017 | Step 3.7 Flash |
|---|---|---|
| 1M in / 250K out | $0.400 | $0.487 |
| 10M in / 2.5M out | $4.00 | $4.88 |
| 100M in / 25M out | $40.00 | $48.75 |
Which should you pick?
Pick DeepSeek V4.1 Flash 1017 when you need the longer context window (1,000,000 vs 262,144 tokens). Pick DeepSeek V4.1 Flash 1017 when price per token dominates your workload — high-volume classification, extraction, or batch summarisation.