GLM 5.3 cut across 2 hosts, by up to 10% at Sail Research (all rates), with cache read down 27% at Wafer
Published 16 September 2026
Price move, per 1M tokens
- Wafer · inputunchanged$0.95 → $0.95
- Wafer · outputunchanged$4.40 → $4.40
- Wafer · cache read↓ 27%$0.26 → $0.19
- Sail Research · input↓ 10%$1.26 → $1.13
- Sail Research · output↓ 10%$3.95 → $3.56
- Sail Research · cache read↓ 10%$0.233 → $0.210
| host | rate | was | now | change |
|---|---|---|---|---|
| Wafer | input | $0.95 | $0.95 | — |
| Wafer | output | $4.40 | $4.40 | — |
| Wafer | cache read | $0.26 | $0.19 | ↓ 27% |
| Sail Research | input | $1.26 | $1.13 | ↓ 10% |
| Sail Research | output | $3.95 | $3.56 | ↓ 10% |
| Sail Research | cache read | $0.233 | $0.210 | ↓ 10% |
Wafer cache read: ↓ 27%Sail Research all rates: ↓ 10%
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 16 September 2026.
read the original at openrouter.ai ↗