Ask HN: Former grok-code-fast-1 users, what coding model are you using now?

2 points by whycombinetor 40 days ago | 3 comments

I get good, cheap, fast feature coding success with grok-4.1-fast for planning and grok-code-fast-1 for execution. But according to the Openrouter usage stats, grok-code-fast-1 is now old hat - usage dropped off a cliff in mid-Feb.

What are you using now? (That's around the same price per token) (AND also blazing fast (100+tps)? is that possible?)

(I'm finding Minimax M2.5 usable, but I think I still prefer grok-code-fast-1. Opus rocks but is expensive - good to have in the quiver for planning a big refactor.)

sgm_ro 39 days ago |

Those groks are quite old by now, and not impressive. Minimax M2.7 is a big upgrade vs 2.5, at a similar price, should easily beat the grok. GLM 5.1 is also a great value (with coding plans), around claude sonnet capability.

But if you are doing a lot of coding your best value will be a coding plan, not pay per token. Gpt plus is a really good value and its 5.4-high model is super capable, some even consider it better than opus.

I don't think it's worth getting less than state of the art models to do actual coding, if the code will actually be used. Costs a bit more but will save you a lot of time in the end, and will produce a better quality product.

whycombinetor 39 days ago | |

For real work I'm already on Opus full time (plan+execute). I think there's a different space for cheap-but-good-enough options for small-to-med repo hobby projects. (Not enough usage to warrant a plan.) And Opus is only ~30 tok/s while grok-code-fast-1 is ~120. So planning with Opus and executing with grok-code-fast-1 is still on the table for me - it's good at following explicitly written plans, lightning fast.

faangguyindia 39 days ago |

Pay for codex and copilot.

It will be much cheaper and if you want to experiment than minimax model