DeepSeek V4 Flash repriced across 5 hosts: Sail Research (US region) input down 16%, NextBit output up 252%
Published 28 September 2026
Price move, per 1M tokens
- NextBit · input↑ 135%$0.150 → $0.352
- NextBit · output↑ 252%$0.300 → $1.056
- NextBit · cache read↓ 66%$0.035 → $0.012
- Wafer · input↑ 106%$0.0413 → $0.0850
- Wafer · outputunchanged$0.35 → $0.35
- Wafer · cache read↑ 43%$0.037 → $0.053
- AtlasCloud · input↑ 13%$0.140 → $0.158
- AtlasCloud · output↑ 70%$0.280 → $0.475
- AtlasCloud · cache read↓ 64%$0.028 → $0.010
- Inceptron · input↑ 67%$0.039 → $0.065
- Inceptron · output↑ 40%$0.271 → $0.380
- Inceptron · cache read↓ 10%$0.030 → $0.027
- Sail Research (US region) · input↓ 16%$0.0225 → $0.0190
- Sail Research (US region) · outputunchanged$0.42 → $0.42
- Sail Research (US region) · cache readunchanged$0.012 → $0.012
- Sail Research · input↓ 12%$0.0215 → $0.0190
- Sail Research · outputunchanged$0.30 → $0.30
- Sail Research · cache readunchanged$0.014 → $0.014
| host | rate | was | now | change |
|---|---|---|---|---|
| NextBit | input | $0.150 | $0.352 | ↑ 135% |
| NextBit | output | $0.300 | $1.056 | ↑ 252% |
| NextBit | cache read | $0.035 | $0.012 | ↓ 66% |
| Wafer | input | $0.0413 | $0.0850 | ↑ 106% |
| Wafer | output | $0.35 | $0.35 | — |
| Wafer | cache read | $0.037 | $0.053 | ↑ 43% |
| AtlasCloud | input | $0.140 | $0.158 | ↑ 13% |
| AtlasCloud | output | $0.280 | $0.475 | ↑ 70% |
| AtlasCloud | cache read | $0.028 | $0.010 | ↓ 64% |
| Inceptron | input | $0.039 | $0.065 | ↑ 67% |
| Inceptron | output | $0.271 | $0.380 | ↑ 40% |
| Inceptron | cache read | $0.030 | $0.027 | ↓ 10% |
| Sail Research (US region) | input | $0.0225 | $0.0190 | ↓ 16% |
| Sail Research (US region) | output | $0.42 | $0.42 | — |
| Sail Research (US region) | cache read | $0.012 | $0.012 | — |
| Sail Research | input | $0.0215 | $0.0190 | ↓ 12% |
| Sail Research | output | $0.30 | $0.30 | — |
| Sail Research | cache read | $0.014 | $0.014 | — |
NextBit output: ↑ 252%NextBit cache read: ↓ 66%Wafer input: ↑ 106%AtlasCloud output: ↑ 70%AtlasCloud cache read: ↓ 64%Inceptron input: ↑ 67%Inceptron cache read: ↓ 10%Sail Research (US region) input: ↓ 16%Sail Research input: ↓ 12%
If you buy
NextBit now suits cache-heavy, input-light workloads, so re-check your output and cache-read mix before your next top-up there.
The model
DeepSeek V4 Flashopen the model →
Open weights290.9B parameters1049K context33 hosts
intelligence65th of 168 writing57th of 168 coding64th of 168 agents20th of 55
DeepSeek V4 Flash is a downloadable text model with a permissive licence, built for long-document work where the bill matters more than peak quality. Its measured quality is weak on the LiveBench boards and mid-pack on the Arena boards, so treat it as a cheap workhorse rather than a quality leader.
Source
Read on openrouter.ai · we published this on 28 September 2026.
read the original at openrouter.ai ↗