gpt-6IMAGE
Kimi K3.1 Pro: 65.7% on Terminal-Bench 4.0, ahead of GPT-6 Astra
Moonshot calls K3.1 Pro its most capable 3T-class open-weight model with a 1M-token context; it scores 65.7% on Terminal-Bench 4.0 versus GPT-6 Astra's 58.2%, 0.7 points behind Claude Opus 5.5.
NeilXbt
@neil_xbtArchived 2 days agoOriginally posted 2026.10.25
NeilXbt @neil_xbtOriginally posted 2026.10.25
Kimi calls K3.1 Pro “our most capable 3T-class open-weight model to date, with a 1-million-token context window.” It scores 65.7% on Terminal-Bench 4.0, ahead of GPT-6 Astra at 58.2% and 0.7 points behind Claude Opus 5.5. Kimi K3.1 Pro is the most stacked open weight release right now!
This post quotes another post: x.com/kimi_moonshot/status/2113681863109939271
x.com/neil_xbt/status/2114268726312726840Archived 2026.10.26 10:10