Alibaba's Qwen3.8-Omni-Flash Counters Google's Gemini 3.8 Live Within Three Days, Cutting Audio Pricing by Over 98%

On September 18, 2026, Alibaba released Qwen3.8-Omni-Flash, a native omnimodal model that supports text, image, audio, and video inputs. Its API audio input pricing is more than 98% lower than that of Google's Gemini 3.8 Live, video task token consumption is cut by 45.7%, and it posts an average improvement of 25% across 29 benchmarks.

Factual Recap

Google released Gemini 3.8 Live on September 15, 2026, taking the lead on speech-to-speech leaderboards. Three days later, Alibaba launched Qwen3.8-Omni-Flash, competing directly on audio pricing. The model supports a 1 million token context, and its video task token consumption drops 45.7% versus its predecessor.

Mechanism Breakdown

Price competition is the core tactic in this round. By sharply lowering audio input pricing, Alibaba is trying to build an advantage at the cost level. Google had previously held the lead in voice interaction with Gemini 3.8 Live, and Alibaba chose to counter on price within a very short window, reflecting a rapid response to the multimodal voice AI market.

The 45.7% reduction in video task token consumption shows progress in efficiency optimization. The average 25% improvement across 29 benchmarks provides a quantitative basis for the performance gains.

Industry Impact

The release could intensify price competition in the multimodal AI sector between China and the United States. The sharp drop in audio pricing helps lower the barrier for developers to integrate the model, but it also puts pressure on other vendors' profit models. Support for a 1 million token context may drive development of long-context application scenarios.

Strategic Assessment

[Analysis] A price war may shift market share in the short term, but long-term sustainability depends on the model's actual performance and ecosystem building. The launch of Qwen3.8-Omni-Flash reflects how Chinese companies are accelerating their pursuit in the multimodal field through cost strategies. Future evaluations need to focus on how it performs in real-world scenarios to judge whether it can sustain its competitiveness.