Artificial Intelligence
Qwen3.8-Omni-Flash nears Gemini Flash benchmarks at lower cost
Qwen said its new multimodal model for AI agents handled audio and video together, used tools independently, and was priced below Gemini 3.8 Flash on cited benchmarks.
- Qwen3.8-Omni-Flash was described as Qwen’s first multimodal model for AI agents.
- Qwen said the model processed audio and video together, used tools on its own, and had a one million token context window.
- Qwen put API pricing at $0.15 per million input tokens and $0.47 per million output tokens.
- Qwen said Gemini 3.8 Flash charged $0.75 per million input tokens and $3.75 per million output tokens at its introductory rate, with those rates due to double on 1 January 2027.
Qwen3.8-Omni-Flash was presented as Qwen’s first multimodal model for agent use. Qwen said it handled audio and video together, drew conclusions, and could use tools to edit vlogs, translate short videos, or summarise films. It also had a one million token context window.
On audio-video tasks, Qwen said the model came close to Gemini 3.8 Flash. Qwen listed API pricing at $0.15 per million input tokens and $0.47 per million output tokens, and said audio input cost under $0.01 per hour.
Qwen estimated 720p video with audio at one frame per second at about $0.20, excluding response costs. It said Gemini 3.8 Flash charged $0.75 per million input tokens and $3.75 per million output tokens at its introductory rate, with those prices due to double on 1 January 2027.
Qwen said the model was available through Qwen Studio, Qwen Cloud, and the API. It also said the open-source Qwen-MM-Plugins added video editing, speaker recognition, PDF video notes, and reusable workflows to Claude Code, Gemini CLI, and Qwen Code, while Qwen-Live Harness supported real-time interaction with a camera and microphone.
Named in this story
Companies
- was named in the pricing comparison for Gemini Flash
Organisations
- Qwen
- said the model was its first multimodal model for AI agents
Products and systems
- Qwen3.8-Omni-Flash
- was the multimodal model discussed
- Gemini 3.8 Flash
- was the benchmarked rival in the comparison
- Qwen Studio
- was one of the services offering the model
- Qwen Cloud
- was one of the services offering the model
- Qwen-MM-Plugins
- were the open-source plugins used with agent tools
- Claude Code
- was one of the tools extended by the plugins
- Gemini CLI
- was one of the tools extended by the plugins
- Qwen Code
- was one of the tools extended by the plugins
- Qwen-Live Harness
- enabled real-time interaction with camera and microphone input
How the source tells it
The piece read as a product announcement, with novelty and price-comparison framing driven largely by Qwen’s own claims.
- novelty the model was introduced as the first multimodal version for agent use
- vendor boosterism capability and pricing comparisons were presented chiefly through the vendor’s own statements