ToolsNov 18, 2026 · 01:371 source

GitHub Copilot Fusion Max Ranks No. 1 on Terminal-Bench 4.1

From marketbrief.now

Image: marketbrief.now
AI;DR

GitHub's Copilot Fusion Max reached a 71.2% resolution rate on the Terminal-Bench 4.1 benchmark, moving it to the top of the leaderboard for coding agents. The score is 5.4 percentage points higher than those of competing frontier models Opus 5.5 and Fable 5.2.

The new system is a fusion of multiple frontier models coordinated by an agent harness. GitHub stated that Copilot Fusion Max costs 88% less than Fable 5.2 and 57% less than Opus 5.5, and is now available across GitHub Copilot product experiences.

Read the sourceSign in to suggest

Sources

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today