Leonard retweeted
Replying to @teortaxesTex
the argument is "claude is by far the best model and so the government will be forced to use it to strengthen the NSA and sow dissent amongst the chinese populace who will threaten a revolt and xi will be forced to capitulate which will weaken china's position on the world stage and the world will all use claude and everything will be safe and aligned and automated and center-left american and any threat to this new world order will be detected and vanquished by claude himself"
Leonard retweeted
Replying to @osanseviero @OfficialLoganK
He’s going to say ‘DolphinGemma’ or some shit isn’t he
Leonard retweeted
Replying to @jasondeanlee
5.1 is the most conscientious, reliable and diligent model out there.
3 has more raw intelligence / understanding but is unreliable and lazy.
A conscientious 3 Deep Think should be amazing.
Leonard retweeted
Anyone else experiencing issues with ChatGPT on the web? Every time I select GPT-5 Pro, it defaults back to GPT-5 Auto instead of staying on my chosen model.
Hey @OpenAI, is this intentional or just a bug? Thanks!
Leonard retweeted
First results are out for the @RepoPrompt benchmark!
Repo Bench is a test set designed to push models on instruction following, large context reasoning, and precision file editing.
Gearing up to release this shortly in the next update so you can run the bench yourself
Leonard retweeted
GPT-5 leaks are suggesting that it can bring users to completion 37% faster than xAI's Waifu Ani on GoonBench.
Leonard retweeted
honestly i would much rather have an open-source 4.1-mini than an open-source o3-mini


















