Skip to main content
AI Choice Engine

Qwen

Qwen 3.8 Omni Flash

Qwen 3.8 Omni Flash is a Qwen model with 1M context, $0.15 in / $0.47 out per million tokens, last verified October 25, 2026. Open weights: no. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

Qwen 3.8 Omni Flash is Alibaba's low-cost omni-modal API model — text, image, audio and video input with a 1M-token context.

Recorded profile verification October 25, 2026Individual provider fields may change

AI model specification details
SpecificationQwen 3.8 Omni Flash
ProviderQwen
TierBudget
Context window1M
Max output131K
Input / 1M tokens$0.15
Output / 1M tokens$0.47
WeightsClosed
ParametersNot verifiedUnverified
Reasoning levelslow, medium, xhigh
Modalitiestext, image, audio, video
API model idqwen3.8-omni-flash
ReleasedSeptember 18, 2026

Pricing tiers: Model Studio International Singapore USD/MTok input/output $0.15/$0.47, implicit cache $0.016. Beijing and Global scopes (Hong Kong, Tokyo, Frankfurt and Virginia) $0.113/$0.382, cache $0.014. The exact non-realtime SKU table gives one token tariff, not a separate verified audio-output rate. Text output only; 1M context, input 991,808 non-thinking/983,616 thinking and output 131,072. The September 21 realtime service is a separate audio-output endpoint.

Best for

  • Hosted multimodal analysis
  • Audio and video understanding
  • Text summaries of recordings

Watch out

Accepts text/image/audio/video and outputs text; audio generation needs a different endpoint. Thinking defaults xhigh; low/medium/xhigh are distinct, minimal maps low, high/max map xhigh, none disables thinking. reasoning_effort with thinking_budget errors. Region-specific keys and workspace endpoints apply; use_multichannel enables spatial audio. The exact card supports implicit and Responses Session caching. Weight release, parameter count and independent benchmark observations remain unverified.

History

Qwen 3.8 Omni Flash timeline

Release and status events for this model that passed two-source verification — dated, sourced, and linked.

  • Qwen3.8-Omni-FlashReleaseSeptember 18, 2026

    Alibaba adds Qwen3.8-Omni-Flash to Model Studio: text, image, audio and video input with a 1M-token context at $0.15/$0.47 per MTok. A -realtime variant followed on 2026-09-21.

    QwenCloud model changelog

Common questions

Qwen 3.8 Omni Flash

Answered from the verified figures on this page rather than general guidance.

How much does Qwen 3.8 Omni Flash cost per million tokens?
Qwen 3.8 Omni Flash lists base rates of $0.15 per million input tokens and $0.47 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. Model Studio International Singapore USD/MTok input/output $0.15/$0.47, implicit cache $0.016. Beijing and Global scopes (Hong Kong, Tokyo, Frankfurt and Virginia) $0.113/$0.382, cache $0.014. The exact non-realtime SKU table gives one token tariff, not a separate verified audio-output rate. Text output only; 1M context, input 991,808 non-thinking/983,616 thinking and output 131,072. The September 21 realtime service is a separate audio-output endpoint.
What is Qwen 3.8 Omni Flash's context window?
Qwen 3.8 Omni Flash accepts about 1M tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Qwen 3.8 Omni Flash best for?
Qwen 3.8 Omni Flash is a budget tier from Qwen. It suits hosted multimodal analysis, audio and video understanding, text summaries of recordings. Accepts text/image/audio/video and outputs text; audio generation needs a different endpoint. Thinking defaults xhigh; low/medium/xhigh are distinct, minimal maps low, high/max map xhigh, none disables thinking. reasoning_effort with thinking_budget errors. Region-specific keys and workspace endpoints apply; use_multichannel enables spatial audio. The exact card supports implicit and Responses Session caching. Weight release, parameter count and independent benchmark observations remain unverified.