Post History

Current version by Nick Antonaccio

Current VersionMay 09, 2026 at 14:11

The prefill (prompt processing) speed is dramatically better on the GX10, compared to the Strix Halo. This can have a significant impact on the overall performance of agentic workflows, especially the startup time required to start a new session. It doesn't make as much of a difference in chat, and for incrementally parsed prompts in a longer session, but when you're dealing with agents that send huge default prompt contexts, or when you're processing large input documents, the GX10 gets cranking much faster. Pi's smaller default prompt contexts can really help improve this performance on the Strix Halo.

Previous Versions
Version 1May 09, 2026 at 14:11

The prefill (prompt processing) speed is dramatically better on the GX10, compared to the Strix Halo. This has a significant impact on the overall performance of agentic workflows. It doesn't make as much of a difference in chat, but boy does Pi work faster with he GX10.