Alibaba · Efficient · Open weights
Qwen 3.8 27B
A 27B dense vision-language model with open weights, sized for a single GPU and focused on coding and office tasks. Also sold as a hosted API on Model Studio.
Model ID: qwen3.8-27b
What that means
A worked example
Per-token pricing is hard to feel. Take a moderately chatty production workload: 100 calls a day, roughly 10,000 tokens in and 2,000 out each time. On Qwen 3.8 27B that comes to about:
$33 / month
Before caching, batching or volume discounts, all of which move this number a long way. Illustrative only.
Strong at
- -Open weights
- -Coding
- -Runs on modest hardware
Typical use
- -Local development
- -Privacy-constrained work
Same tier
What else to look at
| Model | Developer | Context | In / 1M | Out / 1M |
|---|---|---|---|---|
| Claude Haiku 5.5 | Anthropic | 1M | $0.10 | $0.50 |
| Claude Haiku 4.5 | Anthropic | 200K | $1 | $5 |
| GPT-6 Luna | OpenAI | 1.05M | $0.10 | $0.50 |
| GPT-5.6 Luna | OpenAI | 1.05M | $0.20 | $1.20 |
| Gemini 3.5 Flash-Lite | 1.05M | $0.30 | $2.50 |
Verified 2026-10-11.