MiMo
Xiaomi's family of open-weight large language models, offered in Pro and Flash variants.
Liked as a cheap open model for coding work in agents and harnesses like OpenCode,1234 with the Pro variant singled out for strong reasoning.567 The main complaints are slow responses and overthinking,89 and some find it uneven at following instructions.1011
Why people recommend it
Low cost · 3
Commenters say it gets a lot of work done for little money, with near-free cached input tokens.
“cached inputs are practically free”
“You'll get 10x more value by using MiMo 2.5 or DeepSeek Flash which is what I use for most code.”
Show 1 more
“We switched to Mimo 2.6, which is relatively better.”
Coding capability · 3
Several use it as a daily coding model in place of frontier models, including on large codebases.
“Zed + OpenCode + Mimo/GLM/Kimi models are incredible. I haven't touched a "frontier model" for months.”
“I use DeepSeek V4 Flash (high) and MiMo 2.5 (non Pro, because vision)”
Show 1 more
“perfectly capable of following my spec and then poking holes in the implementation”
Pro reasoning · 3
The Pro variant is described as very capable, with better intelligence than DeepSeek Flash for agent work.
“Mimo 2.6 Pro is even better.”
Show 1 more
“I use Pro only for one-off tasks that require complex reasoning”
Concise output · 3
Users note it keeps context tight and is less chatty, with a refreshing communication style.
“Mimo has been one of those models that I have been rooting for since the first I used it”
“Prefer the latter over DeepSeek v4 Pro because it costs the same while being equally good & less chattier for coding workloads.”
Show 1 more
“I really liked it, its communication style was a breath of fresh air compared to how stale gpt-sol is.”
Training transparency · 2
Commenters praise Xiaomi's openness about training methodology, benchmarks and release notes.
“I really like how transparent they’ve been about the training of this model.”
“I love the contrast with yesterday's open-source MiMo release”
Fewer refusals · 1
One team switched to it from GLM because it censored fewer bulk data-processing requests.
“We switched to Mimo 2.6, which is relatively better.”
Caveats and dissent
Slow, overthinks · 2
Users report long time-to-first-token, lengthy chain-of-thought and thinking loops.
“I've personally experienced overthinking and the thinking loop issue more often than seeing it not think enough”
Uneven quality · 3
Some find it uneven or poor at following instructions, and the Flash variant weaker than Pro.
“All behave dumb, don't follow instructions properly even when they write skills for themselves.”
“You should also test MiMo v2.6.”
Show 1 more
Accepts regressions · 1
In one comparison it pushed the user to accept a functional regression instead of fixing it.
“Deepseek, Kimi, MiMo started giving me multiple options but advocating strongly that I should either accept this deficiency”
Data guarantees · 1
One commenter questions the data-training guarantees of the available hosting options for company use.
“Are people using MiMo models as their daily drivers in a company setting?”
Confusing plans · 1
Its token plans are described as a good deal but confusing.
“MiMo has token plans that are a pretty good deal (though confusing”
Pro tool use · 1
One user found the non-Pro model handled tool usage better than Pro.
“Mimo v2.5 non-pro seems to do better with tool usage than its Pro sibling, is much cheaper and solves 90% of the same problems.”
Often compared with
1 moved from GLM to MiMo
1 moved from DeepSeek V4 Flash to MiMo
1 moved from Claude Opus to MiMo
Who recommends it in their own words
Threads where it came up
- MiMo v2.6 5 mentions
- GPT-6 Sol and Luna 3 mentions
- Artificial Analysis Intelligence Index v5 2 mentions
- Claude Opus 5.5 2 mentions
- Qwen 3.6 27B is the sweet spot for local development 2 mentions
- MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second 2 mentions
Also written as MiMo 2.5, MiMo v2.5 Pro, MiMo 2.6, MiMo V2.5 Pro. Data: mimo.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 42
“I've personally experienced overthinking and the thinking loop issue more often than seeing it not think enough”
“I really liked it, its communication style was a breath of fresh air compared to how stale gpt-sol is.”
“I am entirely satisfied when using my Opencode Go plan with open models like Deepseek Flash/Pro, Kimi and MiMo.”
“a little bit of MiMo-2.6”
“You should also test MiMo v2.6.”
“all Mino models are painfully slow”
“MiMo2.6-Pro is a beast”
“Mimo 2.6 Pro is even better.”
“along with some MiMo and DeepSeek”
“So far so good, seem to have done a lot of work for the $2.85 I've spent”
“I use Chinese models on those. GLM, DeepSeek, MiMo.”
“GLM 5.2/5.3, and MiMo thrown in”
“perfectly capable of following my spec and then poking holes in the implementation”
“DS4/MS13/MiMo are almost there for these use cases as well.”
“GLM 5.2/5.3, and MiMo thrown in”
“We switched to Mimo 2.6, which is relatively better.”
“I love the contrast with yesterday's open-source MiMo release”
“flash models, DeepSeek, MiMo, GLM, I love those things”
“I asked MiMo-V2.6-Flash to fork and modify it in a straightforward way.”
“it sounds greate. I still have 50% of my quota this month, so I'll continue renewing to give it a try.”
“Are people using MiMo models as their daily drivers in a company setting?”
“Mimo has been one of those models that I have been rooting for since the first I used it”
“Averages ~25-35tok/s which isn't bad for a first attempt.”
“I really like how transparent they’ve been about the training of this model.”
“mimo-v2.5-free, mimo-v2.5, deepseek-flash in that order”
“MiMo did that a lot in the past”
“MiMo is the dumbest of them all but good enough for basic stuff.”
“Zed + OpenCode + Mimo/GLM/Kimi models are incredible. I haven't touched a "frontier model" for months.”
“mostly GLM-5.3, DS-V4-Flash, and MiMo-2.5-pro”
“DS and MiMo feel very snappy now”
“They are criminally cheap in a way that I don't care to spend tokens.”
“All behave dumb, don't follow instructions properly even when they write skills for themselves.”
“MiMo 2.5 Pro for reasoning”
“I've found this to be true for 1M context-size open weights that I use (GLM 5.2, MiMo v2.5 Pro, & DeepSeek v4 Pro)”
“the only currently really free offering is Mimo via their own CLI”
“Prefer the latter over DeepSeek v4 Pro because it costs the same while being equally good & less chattier for coding workloads.”
“are genuinely good enough as daily drivers, given the costs”
“MiMo has the cheapest tokens for a 1T+ model in the game”
“MiMo has token plans that are a pretty good deal (though confusing”
“I was doing it just fine on llama-server with 200k+ of context.”
“You'll get 10x more value by using MiMo 2.5 or DeepSeek Flash which is what I use for most code.”
“cached inputs are practically free”