Qwen3.8 27B
by Kosmik Compute · family qwen · listed Aug 2026RTVW
Dense 27B vision-language model for coding, agent tasks, and image understanding
Current rates USD per 1M tokens
Input$0.35
Output$2.20
Cache read$0.09
Cache write—
Against the market
On the record
- Model id
qwen/qwen3.8-27b- Context window
- 262K tokens
- Max output
- 33K tokens
- Input modalities
- text, image
- Output modalities
- text
- Reasoning
- yes
- Tool calls
- yes
- Open weights
- yes
- Knowledge cutoff
- —
- Released
- 2026-08-14
- Last updated
- 2026-08-14
- Provider docs
- api.koscompute.com/docs/
Back of the envelope
≈
Also listed at
- Hetznerfree in · free out
- Vercel AI Gateway$0.10 in · $0.40 out
- RunInfra$0.10 in · $0.40 out
- NanoGPT$0.20 in · $1.40 out
- Cortecs$0.334 in · $2.45 out
- Deep Infra$0.40 in · $3.00 out
- Eden AI$0.40 in · $3.00 out
- Weights & Biases$0.40 in · $3.00 out
- Cloudflare Workers AI$0.45 in · $3.20 out
- OpenRouter$0.45 in · $3.20 out
- Kilo Gateway$0.45 in · $3.20 out