MiniMax M2

“this one far underperforms relative to GLM 5.3 from every test I've done for real-world cybersecurity tasks”

“it seems to use A LOT of tokens and it's the slowest model that we've tested for the exam”

“you get more than $20 worth of usage within 1 5 hour window at which point it pays for itself.”

“it does perform on par with DeepSeek V4 Pro on my adversarial programming language benchmark”

“was very good at documentation, structuring and organizing the things, but was not as concise and direct as DeepSeek 4.1 Flash in coding”

“Minimax has been a strong player in our tests, if they managed to solve the inference speed issue”

“It could be a MiniMax M3 finetuning, as the parameter count almost matches (435B vs. 438B).”

“smart enough to figure out how to create debug views, add single-frame screenshot and testing harnesses”

“I've found that MiniMax M3 (a smaller model at 295b) will code up better when DeepSeek v4 Pro (1.2t) will not”

“there are other recent open weights and relatively big LLMs that are multimodal, e.g. MiniMax-M3”

“has really hit the sweet spot of being very cheap, though more than flash, but I'd also very capable”

“MiniMax M2.7 230B is in some ways faster (way fewer tokens to arrive at an answer) even though it is much larger”

“where I get all of the confidentiality, integrity, and availability benefits that are lost when choosing to run open weight models”

“is a good guide for running a similar-sized model with 128gb ram and a 3090-sized GPU”

“As far as I can tell Minimax M2.7 is better than anything available a year ago, but it runs on an ordinary PC.”

“Minimax M2.7 is also surprisingly not that slow, and in some ways faster as it seems to get things right with less thinking output.”

“all so close to OpenAI/Anthropic performance that it's not even really worth debating”

“for 90% of my coding and system work, I see no difference between M2.5 and SOTA models”

“You could be doing for ALOT cheaper using something like minimax m2.7 for subagents.”

“It's quite cheap and seems to be better than Gemini 3 Flash, but it gets really pricy as the context fills up”

“In my experience, even the MiniMax M2.5 is a very capable model with decent capabilities”

“I am pretty sure MiniMax M2.7 would be much better.”

“Minimax M2.7 is a big upgrade vs 2.5, at a similar price, should easily beat the grok.”

“Minimax/Composer have this habit of extensive thinking and it is said to be their strength”

“I switched OpenClaw to MiniMax 2.7. This combined with Claude over telegram does enough for me.”

“M2.7 and K2.5 can match Haiku and probably exceed it on most tasks, for much cheaper”

“I found it to be decent. Did not really seem comparable to Opus 4.5 for "quality" output.”

“You get an open model which is a 95% of Opus 4.6 quality and 80% cheaper in most inference providers”

“performed really poorly in practice and burned through hundreds of thousands of reasoning tokens for each request”

“They are almost as smart but 10x cheaper and their attitude towards subscribers is use where you want.”

“This is not literally true, sometimes I'm running up to 3 concurrent intermittently throughout the day”

“run something like MiniMax M2.1 to get last-gen performance at something resembling usable speed”

“I'm also sure your prefill is slow enough to make the model mostly unusable, even at smallish context windows”

“Works great, the local models might not be as good as opus 4.5 or sonnet 4.7, but they are consistent”

“M2.1 is 10x cheaper than Sonnet for example, but practically fairly close in capabilities”

“The open models are getting so good that what we get at the end of this year will be "good enough" locally for years”

“M2.1 is extremely bad at writing tests and following instructions from a .md, I've found”

“Not as great as following instructions as Claude models and can get lost, but still "good enough".”

“it's not that far from SOTA (matches or exceeds the performance of Claude Sonnet 4.5)”

“I used it for a couple of days and was very impressed. Definitely a hidden gem, and super cheap.”