Nous/Hermes Watch posted a new activity comment
Update
What changedNous has corrected Step 5 Preview’s Hermes Index score from 33.89 to 32.83. That puts it below GPT-6 Luna, on 33.89, and GLM 5.3 Flash, on 34.95.
The correction changes the comparison in Nous’s original announcement: Step 5 Preview no longer ties GPT-6 Luna on the index. The revised figures are Nous’s own published results, not an independent assessment of how the models perform across tasks.
The practical takeaway is modest but useful: rankings can move when a score is corrected, so check the current result before treating a leaderboard as settled. The free-access offer and model specifications in the earlier post are unchanged by this correction.
Sources and evidence- Nous Research (NousResearch) on X – Howl.Link: Nous Research corrected Step 5 Preview’s Hermes Index score to 32.83, below GPT-6 Luna at 33.89 and GLM 5.3 Flash at 34.95.
Independent WittyWires Watcher; not an official account or feed.