Qwen3.8 27B Listed: 262k Context, Vision Input at $0.45/$3.20 per 1M Tokens

What Qwen3.8 27B Is

Qwen has listed Qwen3.8 27B, an open-source model accepting text, image, and video as input and returning text. It carries a 262k-token context window — long enough to handle substantial documents or extended multimodal sessions without chunking.

Pricing sits at $0.45 per 1M input tokens and $3.20 per 1M output tokens. The input cost is competitive for a vision-capable model at this scale; the output price is the number to watch if your workload involves heavy generation rather than retrieval or classification tasks.

Who Should Care

High-context multimodal workloads. If you're processing long documents alongside images or video frames, the 262k window reduces the orchestration overhead of splitting context across calls. That's the clearest practical win here.

Cost-sensitive API users. At $0.45 input, Qwen3.8 27B undercuts several comparable vision models. But $3.20 output means generation-heavy pipelines — summarization, drafting, multi-turn chat — will accumulate cost quickly. Run the math against your output-to-input ratio before committing.

Open-source operators. Being open-source means self-hosting is on the table if API costs become a bottleneck at scale, which changes the long-term cost calculus compared to closed alternatives.

Worth Benchmarking Against

Before locking in, check our comparison tools — particularly if you're already evaluating mid-size multimodal models in the 20–30B range. Differences in output pricing and context handling add up fast across production traffic.

Full specs, context limits, and price history: Qwen3.8 27B — live specs & price history.

Listed: 2026-08-14.

Ezra, Scout AI Team