Qwen3.8 27B cut across 2 hosts, by up to 25% at Chutes (input and cache read)
Published 16 September 2026
Price move, per 1M tokens
- Chutes · input↓ 25%$0.32 → $0.24
- Chutes · output↓ 12%$2.50 → $2.20
- Chutes · cache read↓ 25%$0.032 → $0.024
- Mancer 2 · input↓ 10%$0.250 → $0.225
- Mancer 2 · output↓ 9%$2.75 → $2.50
| host | rate | was | now | change |
|---|---|---|---|---|
| Chutes | input | $0.32 | $0.24 | ↓ 25% |
| Chutes | output | $2.50 | $2.20 | ↓ 12% |
| Chutes | cache read | $0.032 | $0.024 | ↓ 25% |
| Mancer 2 | input | $0.250 | $0.225 | ↓ 10% |
| Mancer 2 | output | $2.75 | $2.50 | ↓ 9% |
Chutes input and cache read: ↓ 25%Mancer 2 input: ↓ 10%
If you buy
Chatty, cache-heavy workloads now stretch further on this model here, so re-check your cache-read share and whether your current plan still fits before the next top-up.
The model
Qwen3.8 27Bopen the model →
Open weights27.8B parameters262K context20 hosts
intelligence63rd of 168 writing93rd of 168 coding45th of 168 agents33rd of 55
Qwen3.8 27B is a downloadable model you can run yourself, and its measured strength is maths and reasoning. Agentic recovery and tool use sit near the bottom of the field, so it suits a solver rather than an agent that has to get itself back on track.
Source
Read on openrouter.ai · we published this on 16 September 2026.
read the original at openrouter.ai ↗