Cache Register Open weights lead Terminal-Bench · As of 2026-10-23 Source on GitHub

An open-weight model now leads Terminal-Bench: Kimi K3.1 Pro beats every big US lab's best. Best model released so far by Anthropic, OpenAI, Google and Meta, and by anyone with open weights; Terminal-Bench 4.0, as measured by Artificial Analysis on October 23, 2026.

Open weights lead Terminal-Bench

An open-weight model now leads Terminal-Bench: Kimi K3.1 Pro beats every big US lab's best
Best model released so far by Anthropic, OpenAI, Google and Meta, and by anyone with open weights; Terminal-Bench 4.0, as measured by Artificial Analysis on October 23, 2026.

Source: Artificial Analysis (artificialanalysis.ai), as of 2026-10-23

Source on GitHub ← All receipts