Since openai removed access to 4.5, I am looking for something comparable from any other company.
Personally, I used it when 4o was not good enough. 4.5 was way better at research and doing more complex programming tasks.
What is comparably good in your experience?
Qwen coder for coding, specifically.
For more general stuff, GLM 4.5 is the new hotness these days. Look up Z AI chat. It’s not great at long context, though.
Jamba is a sleeper model outside of coding and benchmarks. It’s amazing at long context (like feeding it papers) and its world knowledge is great.
And there’s always Gemini Pro. It’s just so smart in general understanding of complex contexts, though it’s pretty “deep fried” and TBH I don’t use it for coding.
For local models, I love Nemotron 49B for stem stuff, and it nicely squeezes into a 3090’s vram as an exl3. Haven’t settled on a coding model beyond Qwen3 32B, though some finetunes are interesting.