DeepSeek V4 Flash repriced across 2 hosts: Phala output up 230%, cache read down 60%
Published 19 September 2026
Price move, per 1M tokens
- Phala · input↑ 120%$0.20 → $0.44
- Phala · output↑ 230%$0.40 → $1.32
- Phala · cache read↓ 60%$0.070 → $0.028
- Relace · inputunchanged$0.040 → $0.040
- Relace · outputunchanged$0.080 → $0.080
- Relace · cache read↑ 100%$0.008 → $0.016
| host | rate | was | now | change |
|---|---|---|---|---|
| Phala | input | $0.20 | $0.44 | ↑ 120% |
| Phala | output | $0.40 | $1.32 | ↑ 230% |
| Phala | cache read | $0.070 | $0.028 | ↓ 60% |
| Relace | input | $0.040 | $0.040 | — |
| Relace | output | $0.080 | $0.080 | — |
| Relace | cache read | $0.008 | $0.016 | ↑ 100% |
Phala output: ↑ 230%Phala cache read: ↓ 60%Relace cache read: ↑ 100%
If you buy
Phala's output cost now dominates for chat-heavy use, so re-check your prompt and completion mix before your next top-up, while cache-heavy workloads gain.
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 from the hosts' own published rates · we published this on 19 September 2026.