Opus 4.6 Fast Mode – 2.5x better token throughput in Copilot (github.blog)

The much-anticipated Fast Mode for Claude Opus 4.6 has been launched in a research preview on GitHub Copilot, promising token throughput speeds up to 2.5 times faster than previous iterations. Despite this increase in speed, the model maintains the same level of intelligence, making it a significant advancement for developers and AI enthusiasts. This feature is currently being rolled out gradually and is available to Copilot Pro+ and Enterprise users, allowing them to access the enhanced performance across various modes, including chat, ask, edit, and more. The implications of this release are profound for the AI/ML community, particularly in improving workflow efficiency for software development. By significantly reducing the time it takes for Copilot to generate output, developers can expect expedited coding and a smoother integration of AI assistance into their projects. Users will need to enable Fast Mode in the Copilot settings, and the rollout will be gradual, so feedback from the community will be crucial in refining this experimental feature. This development highlights a strong focus on enhancing inference speed while preserving quality, a key concern for many AI applications.