DeepSeek V4.1 Flash repriced across 6 hosts: Io Net input down 64%, Ionstream input up 97%
Published 30 September 2026
Price move, per 1M tokens
- Ionstream · input↑ 97%$0.145 → $0.285
- Ionstream · outputunchanged$1.15 → $1.15
- Ionstream · cache readunchanged$0.005 → $0.005
- InferenceNet · input↓ 42%$0.069 → $0.040
- InferenceNet · output↑ 67%$0.45 → $0.75
- InferenceNet · cache readunchanged$0.020 → $0.020
- Io Net · input↓ 64%$0.250 → $0.090
- Io Net · output↓ 59%$1.00 → $0.41
- Io Net · cache read↓ 70%$0.030 → $0.009
- OpenInference · input↓ 34%$0.0300 → $0.0198
- OpenInference · output↓ 21%$0.500 → $0.396
- OpenInference · cache read↓ 71%$0.0100 → $0.0029
- Wafer · input↓ 25%$0.100 → $0.075
- Wafer · outputunchanged$0.44 → $0.44
- Wafer · cache readunchanged$0.045 → $0.045
- Morph · input↓ 3%$0.0805 → $0.0780
- Morph · output↓ 1%$0.510 → $0.504
- Morph · cache read↑ 270%$0.0027 → $0.0100
| host | rate | was | now | change |
|---|---|---|---|---|
| Ionstream | input | $0.145 | $0.285 | ↑ 97% |
| Ionstream | output | $1.15 | $1.15 | — |
| Ionstream | cache read | $0.005 | $0.005 | — |
| InferenceNet | input | $0.069 | $0.040 | ↓ 42% |
| InferenceNet | output | $0.45 | $0.75 | ↑ 67% |
| InferenceNet | cache read | $0.020 | $0.020 | — |
| Io Net | input | $0.250 | $0.090 | ↓ 64% |
| Io Net | output | $1.00 | $0.41 | ↓ 59% |
| Io Net | cache read | $0.030 | $0.009 | ↓ 70% |
| OpenInference | input | $0.0300 | $0.0198 | ↓ 34% |
| OpenInference | output | $0.500 | $0.396 | ↓ 21% |
| OpenInference | cache read | $0.0100 | $0.0029 | ↓ 71% |
| Wafer | input | $0.100 | $0.075 | ↓ 25% |
| Wafer | output | $0.44 | $0.44 | — |
| Wafer | cache read | $0.045 | $0.045 | — |
| Morph | input | $0.0805 | $0.0780 | ↓ 3% |
| Morph | output | $0.510 | $0.504 | ↓ 1% |
| Morph | cache read | $0.0027 | $0.0100 | ↑ 270% |
Ionstream input: ↑ 97%InferenceNet input: ↓ 42%InferenceNet output: ↑ 67%Io Net input: ↓ 64%OpenInference input: ↓ 34%Wafer input: ↓ 25%Morph input: ↓ 3%Morph cache read: ↑ 270%
If you buy
Ionstream now costs more for the same model, so if you run long prompts there, re-check your spend before the next top-up and consider whether your workload still fits.
The model
DeepSeek V4.1 Flashopen the model →
Open weights763.2B parameters1049K context35 hosts
intelligence21st of 168 writing37th of 168 coding11th of 168 agents13th of 55
DeepSeek V4.1 Flash is a downloadable model with a permissive licence, and it places 11th of 168 on Arena Coding via Max as of 25 Sep 2026. At 763.2 billion parameters it is a data-centre job to run yourself, so hosted use is the practical route for most readers.
Source
Read on openrouter.ai · we published this on 30 September 2026.
read the original at openrouter.ai ↗