Qwen3.8-Omni-Flash Prices Multimodal AI at $0.15/$0.47 per Million Tokens, Undercutting Gemini Flash by 5x

What happened

Alibaba's Qwen team has released Qwen3.8-Omni-Flash, described as the company's first multimodal model built specifically for AI agents. The model processes audio and video simultaneously, draws conclusions from combined inputs, and can autonomously invoke tools to edit vlogs, translate short videos, or summarize movies.

The context window spans 1 million tokens. According to Qwen, the model performs close to Google's Gemini 3.8 Flash on audio-video benchmarks — though the company has not published specific comparative benchmark scores alongside the claim.

Pricing

Qwen3.8-Omni-Flash is priced at $0.15 per million input tokens and $0.47 per million output tokens through the Qwen API. Qwen estimates audio input costs under $0.01 per hour of content, and processing 720p video with audio at one frame per second runs about $0.20, excluding response generation costs.

By comparison, Gemini 3.8 Flash charges $0.75 per million input tokens and $3.75 per million output tokens at its introductory rate — roughly 5x and 8x more than Qwen's rates, respectively. Google's pricing is set to double on January 1, 2027.

Availability and tooling

The model is accessible through Qwen Studio, Qwen Cloud, and the standard API. Alongside the model release, Qwen introduced open-source Qwen-MM-Plugins, which add video editing, speaker recognition, PDF-to-video note generation, and reusable agent workflows. These plugins integrate with existing agent frameworks including Claude Code, Gemini CLI, and Qwen Code.

A separate tool, Qwen-Live Harness, enables real-time interaction using a device's camera and microphone, positioning the release toward live agentic use cases rather than purely batch processing.

What this means

Qwen3.8-Omni-Flash is a direct pricing challenge in the multimodal agent space, where Gemini Flash has held a strong position for cost-sensitive, high-volume audio and video workloads. If Qwen's benchmark parity claims hold up under independent testing, the price gap — roughly 5x cheaper on input, 8x cheaper on output — could pull developers building video-processing agents, vlog editors, and real-time assistants toward Alibaba's stack, particularly in cost-per-hour-of-video use cases like the ~$0.20 video-processing estimate.

The open-sourcing of Qwen-MM-Plugins for third-party agent frameworks (Claude Code, Gemini CLI) is notable: it signals Qwen is optimizing for adoption within existing developer tooling rather than trying to build a closed ecosystem. That said, Qwen's benchmark comparisons are self-reported, and Gemini 3.8 Flash's actual real-world audio-video accuracy under production loads hasn't been independently verified against these figures. Buyers evaluating a switch should run their own comparative tests before committing to either platform's pricing model, especially given Google's scheduled price increase in 2027.