Qwen3.8-Omni-Flash

Qwen3.8-Omni-Flash is Alibaba's first omni-modal model built around agentic work, released September 18, 2026 as qwen3.8-omni-flash. It takes text, image, audio, and video on one endpoint and is built natively on the Qwen3.8-Flash foundation rather than bolting perception encoders onto a text model. QwenCloud lists it at $0.15 per million input tokens and $0.47 output, with implicit cache reads at $0.016, a 1 million token context window and up to 131K output tokens. The headline is not the per-token rate, which matches Qwen3.8-Flash exactly, but what that rate does to media: Alibaba says an hour of audio input costs 98 percent less than on Qwen3.5-Omni-Plus and an hour of combined audio and video costs more than 93 percent less, which is the difference between transcribing a meeting archive as an experiment and doing it as a standing job. It handles up to an hour of continuous audio or audio-video per call, recognizes speech in 74 languages, and Alibaba reports an average gain of more than 25 percent across 29 evaluations against Qwen3.5-Omni-Plus, all vendor-reported. One limitation deserves to be read before you design around it: despite the Omni name and a predecessor that generated speech, this model outputs text only, so anything voice-facing still needs a separate text-to-speech stage. Weights have not been published. Available through Qwen Chat, QwenCloud, and the Model Studio API on an OpenAI-compatible endpoint.