Qwen3.8 Omni Flash
Qwen3.8 Omni Flash is an omni-modal reasoning model from Alibaba and the first Qwen built around agentic capabilities with native audio-video understanding. It accepts text, images, audio and video in one model, with a 1M-token context window - at $0.15 per million input tokens.
Technical Specifications
Live data fetched from the OpenRouter model registry — always up to date.
What is Qwen3.8 Omni Flash?
Qwen3.8 Omni Flash is unusual in two ways at once. It is omni-modal - text, images, audio and video all go in - and it is cheap, at $0.15 per million input tokens. Alibaba positions it as the first Qwen model designed around agentic use with native audio-video understanding, which makes it the practical option for media-heavy pipelines that would be unaffordable on a frontier model.
Features & Capabilities
Audio and Video Input
Native understanding of recorded audio and video, not just frames.
Flash Pricing
$0.15 per million input tokens for omni-modal work.
1M Context
Long recordings and their transcripts in one request.
Agent-First Design
Built around agentic use cases with native audio-video understanding.
Reasoning Model
Reasons over the media rather than only transcribing it.
Media Pipelines at Scale
Cheap enough to run over an entire library rather than a sample.
Who Is Qwen3.8 Omni Flash For?
Media Summarisation
Summarising podcasts, calls and recordings at volume.
Video Content Analysis
Finding moments, topics and claims in long footage.
Meeting Intelligence
Notes and action items from audio plus slides.
Accessibility Work
Descriptions and captions generated from source media.
Qwen3.8 Omni Flash Pricing
- $0.15 per million input tokens, $0.47 per million output tokens
- Omni-modal input at Flash-tier pricing
- Run it from your Hoodgen free credits, no credit card required
- Higher allowances on Hoodgen AI paid plans
What Makes Qwen3.8 Omni Flash Different?
The combination on offer here is what makes Qwen3.8 Omni Flash worth attention: native audio and video understanding in a model that costs $0.15 per million input tokens. Models that accept video generally sit in the frontier price bracket; models this cheap usually stop at text. Omni Flash is both, and that changes what is economically sensible.
It also reasons over the media rather than only transcribing it. You can ask what was decided in a recording, whether a claim in a video matches a written document, or where in a long session a topic first appears. Paired with a 1M-token context window that means a substantial recording plus its transcript can be handled in a single request.
The tradeoff is depth. Flash-tier models are fast and economical rather than exhaustive, so for a genuinely difficult judgement call the frontier tier still wins. The productive pattern is to filter with Omni Flash and escalate the few hard cases - Hoodgen keeps both on the same account.
Qwen3.8 Omni Flash Performance & Pricing
Omni Flash pairs audio and video understanding with Flash-tier pricing, which is rare - media-capable models usually cost many times more.
What Can You Do With Qwen3.8 Omni Flash?
Use Omni Flash wherever media volume would otherwise make the work unaffordable. It turns an entire library of recordings into something you can process rather than sample.
Best Use Cases
- Summarising every call in a month of recordings
- Finding specific moments across a video archive
- Extracting action items from meetings
- Checking media claims against written sources
- Generating captions and descriptions at scale
Qwen3.8 Omni Flash Privacy & Data Policy
Qwen3.8 Omni Flash is developed by Alibaba Qwen team and served through OpenRouter. Prompts and completions are processed under the provider terms.
Because this model accepts audio and video, be deliberate about what you upload. Avoid sending confidential recordings or personal media to any hosted model without checking the provider terms first.
Qwen3.8 Omni Flash vs Other Models
Frequently Asked Questions
Media understanding at Flash prices
Chat with Qwen3.8 Omni Flash on Hoodgen - sign in and start in seconds.
Chat with Qwen3.8 Omni Flash