DeepSeek V4 Pro repriced across 3 hosts: Reka input and cache read down 55%, Relace input and cache read up 36%
Published 1 October 2026
Price move, per 1M tokens
- Reka · input↓ 55%$1.30 → $0.58
- Reka · output↓ 33%$2.60 → $1.74
- Reka · cache read↓ 55%$0.130 → $0.058
- Relace · input↑ 36%$0.123 → $0.167
- Relace · outputunchanged$3.50 → $3.50
- Relace · cache read↑ 36%$0.123 → $0.167
- Wafer · input↑ 15%$0.48 → $0.55
- Wafer · outputunchanged$4.20 → $4.20
- Wafer · cache read↑ 15%$0.384 → $0.440
| host | rate | was | now | change |
|---|---|---|---|---|
| Reka | input | $1.30 | $0.58 | ↓ 55% |
| Reka | output | $2.60 | $1.74 | ↓ 33% |
| Reka | cache read | $0.130 | $0.058 | ↓ 55% |
| Relace | input | $0.123 | $0.167 | ↑ 36% |
| Relace | output | $3.50 | $3.50 | — |
| Relace | cache read | $0.123 | $0.167 | ↑ 36% |
| Wafer | input | $0.48 | $0.55 | ↑ 15% |
| Wafer | output | $4.20 | $4.20 | — |
| Wafer | cache read | $0.384 | $0.440 | ↑ 15% |
Reka input and cache read: ↓ 55%Relace input and cache read: ↑ 36%Wafer input and cache read: ↑ 15%
If you buy
Relace now costs more to run for input-heavy and cache-reliant workloads, so re-check your usage mix and whether this host still fits before your next top-up.
The model
DeepSeek V4 Proopen the model →
Open weights1598.8B parameters1049K context28 hosts
intelligence36th of 168 writing32nd of 168 coding43rd of 168 agents31st of 55
DeepSeek V4 Pro is a large downloadable text model with a one-million-token request limit and strong mathematics scores. Its mixture-of-experts design keeps only 37 billion parameters active per token, making it more efficient to run than its total size suggests.
Source
Read on openrouter.ai · we published this on 1 October 2026.
read the original at openrouter.ai ↗