GLM 5.3 cut across 2 hosts, by up to 43% at Inceptron (output), with cache read down 44% there
Published 26 September 2026
Price move, per 1M tokens
- Inceptron · input↓ 36%$0.597 → $0.380
- Inceptron · output↓ 43%$2.50 → $1.43
- Inceptron · cache read↓ 44%$0.152 → $0.085
- Relace · input↓ 21%$0.70 → $0.55
- Relace · output↓ 23%$2.20 → $1.70
- Relace · cache read↓ 23%$0.13 → $0.10
| host | rate | was | now | change |
|---|---|---|---|---|
| Inceptron | input | $0.597 | $0.380 | ↓ 36% |
| Inceptron | output | $2.50 | $1.43 | ↓ 43% |
| Inceptron | cache read | $0.152 | $0.085 | ↓ 44% |
| Relace | input | $0.70 | $0.55 | ↓ 21% |
| Relace | output | $2.20 | $1.70 | ↓ 23% |
| Relace | cache read | $0.13 | $0.10 | ↓ 23% |
Inceptron output: ↓ 43%Relace output and cache read: ↓ 23%
If you buy
Long-context and cache-heavy work now fits a smaller budget here, so re-check your usage mix and cache settings before your next top-up.
The model
GLM 5.3open the model →
Proprietary1049K context38 hosts
intelligence16th of 168 writing19th of 168 coding23rd of 168 agents17th of 55
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 26 September 2026.
read the original at openrouter.ai ↗