> Models that you can run at home (Like Qwen 35B) aren't remotely close to Opus or GPT 5.5.
replyIs that characterization based on some objective facts or benchmarks?
Yes, there aren't any 35B models that are beating frontier models at just about anything generalized
replyBased on private test prompts I've run through OpenRouter.
replyI don't need a Ferrari to get to work
replyBut you need the best tools to do the job
reply