# Qwen3.8-Omni-Flash

Alibaba, released 2026-09-14

**Verdict: conditional.** Qwen3.8-Omni-Flash takes text, images, audio and video in one call with a 1M-token context, and Alibaba says it cut the per-hour price of audio input by more than 98% against the previous version. Conditional because every benchmark we found is Alibaba's own and the model answers in text only; spoken replies need a separate model, such as the Realtime variant, at higher rates. Right for teams building audio-visual agents on Alibaba Cloud. Last verified 2026-09-23.

## Pricing

- Input: $0.15 / 1M
- Output: $0.47 / 1M (text)
- Context: 1M ctx; cache hit $0.016/1M; International

## Strengths

- Long recordings: up to 3 hours of audio or 2 hours of video per file, and up to 64 files per request
- Multilingual speech input across 113 languages and dialects
- Agents that need thinking, function calling and web search on multimodal input in one call
- Cost-sensitive multimodal input at $0.15 per 1M input tokens, $0.016 on cache hits
- Spoken output: the standard model returns text; voice replies need a separate model, Qwen3.8-Omni-Flash-Realtime or Qwen3.5-Omni
- Self-hosting: we found no open-weights release, only the API and open-source plugins
- Decisions that need independent benchmarks: every score we found is Alibaba's own

## Benchmarks

- WildClawBench-MM (Alibaba-reported): 71.0 (https://qwen.ai/blog?id=qwen3.8-omni-flash)
- AgenticVBench (Alibaba-reported): 36.8 (https://qwen.ai/blog?id=qwen3.8-omni-flash)

## Sources

- [Alibaba Cloud Model Studio pricing](https://www.alibabacloud.com/help/en/model-studio/model-pricing)
- [Qwen blog: Qwen3.8-Omni-Flash launch post](https://qwen.ai/blog?id=qwen3.8-omni-flash)
- [Alibaba Cloud Model Studio: Qwen-Omni models](https://www.alibabacloud.com/help/en/model-studio/qwen-omni)

Canonical: https://neomanex.com/models/qwen3-8-omni-flash