62 points seizethecheese 1 hour ago 16 comments
afavour 45 minutes ago | parent
For this model, I mean. And not the Anthropic IPO.
seizethecheese 41 minutes ago | parent
Just as good but cheaper comes from the benchmark performance. Also faster from openrouter. The opus 5.5 parity projection is an internal model based on those assumptions from the Agent Duel team.
enraged_camel 12 minutes ago | parent
quinndupont 41 minutes ago | parent
seizethecheese 33 minutes ago | parent
hedgehog 25 minutes ago | parent
rootusrootus 35 minutes ago | parent
But that's about as complicated as I get. I have recently tried to use Copilot because that is what work wants us to use, and even with Opus 5.5 it is noticeably inferior to CC on the same model. Wild. So clearly the harness matters.
seizethecheese 29 minutes ago | parent
For what it’s worth, the linked article data (Agent Duel data) is all run on OpenCode. I’m curious if people would prefer we run each model in its native harness (at the expense of user experience because it’ll be buggy).
thefunnyman 26 minutes ago | parent
pritambarhate 24 minutes ago | parent
scottyah 16 minutes ago | parent
- SWE-bench Pro (hard 51): 39.2% → 64.7%
- Terminal-Bench 2.1: 39.5% → 55.8%
anukin 5 minutes ago | parent
bwat49 24 minutes ago | parent
wmf 18 minutes ago | parent
seizethecheese 6 minutes ago | parent
The blog itself can be seen more as an exploration of how speed impacts coding agent battle votes, than anything about Neon.
Applying the speed impact to Gemini Carbon is particularly interesting because it's the top model in our leaderboard when you exclude early votes (https://agentduel.com/leaderboard?excludeEarlyVotes=true) but mediocre when you let developers vote. Gemini literally takes 3x as long as opponents in matched battles.
lacoolj 7 minutes ago | parent
Second, I will be trying this with agy when my Opus 5.5 limit hits. I only hope its faster than Argon which is already dog slow