Qwen3.8-Flash

// Overview

Qwen3.8-Flash is the Flash tier of Alibaba's Qwen3.8 generation, served on Alibaba Cloud Model Studio under the invocation id `qwen3.8-flash` and listed on OpenRouter with a creation date of 26 August 2026. It is the hosted, production counterpart to Qwen3.8-Flash-Next, the open-weight checkpoint the Qwen team published the same day as an architecture preview of Qwen4. Alibaba's documentation describes it as a multimodal model that combines reasoning and generation with high speed, aimed at coding assistance, agentic workflows and visual understanding.

It accepts text, image and video as input and returns text. The context window is 1,000,000 tokens, with a maximum input of 991,808 tokens (983,616 in thinking mode), a maximum output of 131,072 tokens, and a thinking chain of up to 262,144 tokens. Model Studio exposes function calling, structured output, context caching, prefix continuation and the model playground, plus web search in every region except Frankfurt and Tokyo; batch inference and fine-tuning are not supported.

Pricing is flat rather than tiered by prompt length. In the Beijing, Frankfurt, Tokyo and Virginia regions Model Studio lists ¥0.8 per 1M input tokens and ¥2.7 per 1M output tokens, with cache hits at ¥0.1 and cache creation at ¥1.25; Singapore is higher at ¥1.094 / ¥3.427, with cache hits at ¥0.117 and cache creation at ¥1.458. Alibaba Cloud's international documentation quotes the same rates as $0.113 / $0.382 for Beijing and $0.15 / $0.47 for Singapore, and OpenRouter resells the model at $0.15 / $0.47 per 1M tokens. Rate limits are 30,000 requests and 5,000,000 tokens per minute in most regions, and 15,000 / 2,000,000 in Singapore.

// Pricing

Alibaba Cloud Model Studio list price in the Beijing, Frankfurt, Tokyo and Virginia regions; cache creation is ¥1.25. Singapore is ¥1.094 / ¥3.427 with cache hits at ¥0.117 and cache creation at ¥1.458. Alibaba Cloud's international documentation quotes these as $0.113 / $0.382 (Beijing) and $0.15 / $0.47 (Singapore); OpenRouter lists the model at $0.15 / $0.47 per 1M tokens.

Pricing source ↗

// Strengths

• 1,000,000-token context with text, image and video input at the cost floor of the Qwen3.8 line
• Thinking mode with a chain-of-thought budget of up to 262,144 tokens and 131,072 tokens of output
• Flat per-token pricing rather than the prompt-length tiers used by the Qwen3.7 Flash generation
• Function calling, structured output, context caching and prefix continuation available on Model Studio
• Hosted counterpart to the open-weight Qwen3.8-Flash-Next, so a self-hosted checkpoint of the same architecture family exists

// Best for

• Reach for it as the cheap sub-agent or fan-out worker behind a more expensive planner model on high-volume traffic.
• Reach for it for screenshot, document and video understanding pipelines where per-token cost dominates the bill.
• Reach for it for whole-codebase or whole-corpus passes that have to fit inside a single 1M-token request.
• Reach for it for agentic tool-calling loops that need structured output and cached context across many turns.

// How to access

// Qwen-Flash (vision-language) — every version

The full lineage of the Qwen-Flash (vision-language) line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.