Best models per $ in actual development scenarios? Hi all, joined recently. I have seen yesterday the latest model benchmarks as shared by @millions-crayon-32880. Very informative but it didn't cover all the models supported. Also benchmarks can be different from practical tasks performances. I was wondering if you found models way cheaper than Claude and the GPT-5 that still get your job done in a satisfactory manner. What about the Grok models, for instance?

gemini-3-flash might be a good choice, although we haven’t taken a close look at it.