Qwen3.8 27B

• Type: Generation
• Capabilities: reasoning, vision

Overview

Qwen3.8-27B is a compact multimodal reasoning model from Alibaba’s Qwen family, designed for general-purpose reasoning, coding, tool use, and vision workloads. Its 262K-token context window supports long documents and extended agentic tasks, while the FP8 deployment offers efficient serving.

---

Thinking Mode:

This model reasons step-by-step before responding by default.

Set reasoning_effort to none to disable thinking. Other effort levels enable thinking but do not select graduated reasoning budgets.

Pricing

Priority	Input Tokens (per 1M)	Cache Read Tokens (per 1M)	Output Tokens (per 1M)
Realtime1	$0.45	$0.45	$3.00
Async	$0.35	$0.35	$2.25
Batch (24h)	$0.25	$0.25	$1.50

Playground

Open this model in the Playground.

Footnotes

1. Realtime availability is limited. Doubleword is primarily a batch API. ↩