Re:Xiaomi MiMo-v2.5 Series API Permanent Price Reduction Up ...
by zarify on Wednesday May 27, 2026 @05:53AM
(#48291922)
Funny, I had the opposite experience with MiniMax and Mimo when using OpenCode. MiniMax got stuck with looping through broken tool calls all the time and MiMo just powered through things and for the most part just worked.
Re:Xiaomi MiMo-v2.5 Series API Permanent Price Reduction Up ...
by shanoaice on Friday May 29, 2026 @03:48PM
(#48328280)
similarly for me, MiniMax is kind of horrible that it somewhat regularly fall into loops that I had to save it from. DeepSeek & MiMO rarely got stuck. wonder how you get completely reversed experience.
Re:Xiaomi MiMo-v2.5 Series API Permanent Price Reduction Up ...
by _davide_ on Tuesday June 09, 2026 @03:11AM
(#48457652)
I did develop my own agent around MiniMax. I did see weird behavior when I messed up the loop, like omitting pieces of remove thinking; maybe it's an agent bug, some models/providers just ignore/normalize the broken input, some don't.
Re:Xiaomi MiMo-v2.5 Series API Permanent Price Reduction Up ...
by ignoramous on Wednesday June 10, 2026 @03:17AM
(#48472655)
Seems like the harness may have to be "finetuned" for models. I've had my share of MiMo 2.5 Pro get stuck in thinking loops with GitHub Copilot.
MiniMax (currently 2.7) which is a ~270B model tuned exclusively for agentic purposes, performs so MUCH better; it's more reliable and cheaper. Both are still far away from Opus 4.7 that I'm using at work. IMO benchmarks are just a very rough estimation; everyone cheats as much as they can get away with. Test the model yourself; do not make any assumptions based on the benchmarks.
I would love to see specialized, cheaper, bleeding-edge models like MiniMax for other non-agentic purposes as well. Why pay $1 for a general model when, for example, you can pay $0.1 for a content-moderator model that you actually need?