4 ms·

Open-weight 27B hits 38% on Terminal-Bench 2.0 (Opus 4.1 hit 38% in Aug 2025)

• ubermon 5mo ago
  [dead]
• debpack 5mo ago
  this is super sick man
  • ubermon 5mo ago
    Thanks!
• timothyshen123 5mo ago
  Interesting find on this. Thanks for sharing
  • ubermon 5mo ago
    Thank you! I think there is a lot to dive deep later with different hardware, inference engine, prompt/harness setup etc.
• annjose 5mo ago
  > today's best runnable-offline model is roughly 6–8 months behind today's frontier. But it doesn't matter because frontier models were extremely good 8 months ago and we were doing real work with them. Now we have more capable open-source agents like pi and OpenCode which work well with these models. More importantly, offline models is the best choice for privacy, on-device inference and no token/cost anxiety.
  • ubermon 5mo ago
    Totally agree! I think we are very early on discovering the full potential of local models
  • swrrt 5mo ago
    Yep, offline mode is useful for edge devices also. I am considering deploying a extremely small model on steam deck actually.
• merkleforest 5mo ago
  > 2. How local use feels in practice Do we have stats on how does the models do on Mac M-series chips?
  • ubermon 5mo ago
    Not yet, will conduct a more comprehensive one later