GLM 5.3 repriced across 5 hosts: Sail Research (US region) input down 33%, output up 100%
Published 22 October 2026
Price move, per 1M tokens
- Sail Research (US region) · input↓ 33%$0.75 → $0.50
- Sail Research (US region) · output↑ 100%$1.00 → $2.00
- Sail Research (US region) · cache readunchanged$0.080 → $0.080
- Sail Research · input↓ 33%$0.75 → $0.50
- Sail Research · output↑ 100%$1.00 → $2.00
- Sail Research · cache readunchanged$0.080 → $0.080
- Inceptron · input↓ 18%$0.55 → $0.45
- Inceptron · output↑ 40%$1.40 → $1.96
- Inceptron · cache read↓ 33%$0.18 → $0.12
- Fireworks · input↓ 30%$1.40 → $0.98
- Fireworks · output↓ 30%$4.40 → $3.08
- Fireworks · cache read↓ 30%$0.260 → $0.182
- Io Net · inputunchanged$0.49 → $0.49
- Io Net · outputunchanged$3.20 → $3.20
- Io Net · cache read↓ 21%$0.19 → $0.15
- InferenceNet · inputunchanged$0.080 → $0.080
- InferenceNet · output↓ 9%$4.38 → $4.00
- InferenceNet · cache readunchanged$0.076 → $0.076
| host | rate | was | now | change |
|---|---|---|---|---|
| Sail Research (US region) | input | $0.75 | $0.50 | ↓ 33% |
| Sail Research (US region) | output | $1.00 | $2.00 | ↑ 100% |
| Sail Research (US region) | cache read | $0.080 | $0.080 | — |
| Sail Research | input | $0.75 | $0.50 | ↓ 33% |
| Sail Research | output | $1.00 | $2.00 | ↑ 100% |
| Sail Research | cache read | $0.080 | $0.080 | — |
| Inceptron | input | $0.55 | $0.45 | ↓ 18% |
| Inceptron | output | $1.40 | $1.96 | ↑ 40% |
| Inceptron | cache read | $0.18 | $0.12 | ↓ 33% |
| Fireworks | input | $1.40 | $0.98 | ↓ 30% |
| Fireworks | output | $4.40 | $3.08 | ↓ 30% |
| Fireworks | cache read | $0.260 | $0.182 | ↓ 30% |
| Io Net | input | $0.49 | $0.49 | — |
| Io Net | output | $3.20 | $3.20 | — |
| Io Net | cache read | $0.19 | $0.15 | ↓ 21% |
| InferenceNet | input | $0.080 | $0.080 | — |
| InferenceNet | output | $4.38 | $4.00 | ↓ 9% |
| InferenceNet | cache read | $0.076 | $0.076 | — |
Sail Research (US region) input: ↓ 33%Sail Research (US region) output: ↑ 100%Sail Research input: ↓ 33%Sail Research output: ↑ 100%Inceptron input: ↓ 18%Inceptron output: ↑ 40%Fireworks all rates: ↓ 30%Io Net cache read: ↓ 21%InferenceNet output: ↓ 9%
If you buy
Output-heavy work at this host now costs more per reply, so re-check your usage mix before the next top-up and trim long answers if the bill matters.
The model
GLM 5.3open the model →
Proprietary1049K context48 hosts
intelligence19th of 184 writing22nd of 184 coding28th of 184 agents28th of 66
GLM 5.3 is a hosted-only text model: we list no download for it, so using it means choosing a host. It is strongest where people rate the answers themselves, and mid-field on the harder task sets.
Source
Read on openrouter.ai · we published this on 22 October 2026.
read the original at openrouter.ai ↗