I been using MiMo-V2.5 to do most of my work as software engineer, on a variety of projects I'm working on, and I been VERY happy with ROI. The model is very powerful! Not perfect – I've run in hallucination loops once or twice, but nothing a stop-then-continue wouldn't solve.
The cost is unbelievably low, and the quality of intelligence I get is equivalent to when I was working mostly with Anthropic models (late last year/early this year). I'm fully invested in MiMo and I'm very happy with it.
--
PS: I also check almost daily to see if other models are capable of doing such great work. And they do – DS4F is powerful and DS41 is impressive, GLM 5.3 Flash gets a job done well, etc. – but when I add cost of M-token in the ROI math, Jeez! MiMo is an order of magnitude better.
rapind 58 minutes ago [-]
I’ve been very pleased with DS 4.1 flash. Not so much the 4.0 models, but for coding (Rust) it’s been great so far (3 solid days of work).
I’ll give Mimo a try.
trollbridge 35 minutes ago [-]
MiMo is my backup whenever DeepSeek is down, had the price bump, is slow, etc.
UltraSpeed was absolutely awesome. I miss it.
DS 4.1 Flash is amazing. Well worth the extra cost.
flexagoon 3 hours ago [-]
How does it compare with DS 4.1 Flash in your experience, if you ignore the cost?
alwinaugustin 3 hours ago [-]
I am also using 2.5 and it is giving me solid results. Its available free on Openrouter
walrus01 5 hours ago [-]
I've found that mimo v2.5 works for very basic things like a python script to do one thing, but it also is very 'dumb' compared to qwen 3.8-flash-next (I think the benchmark scores for terminal and coding specific benches back this up). And definitely not in the same class as like a GLM5.2 or 5.3. It's fast but makes basic mistakes that only get caught later.
girvo 3 minutes ago [-]
The fact I can run Qwen 3.8 Flash Next locally, forever (on my DGX Spark-alike) is genuinely shocking to me. It’s crazy good for how small it is. Fast, too.
3 hours ago [-]
jwpapi 5 hours ago [-]
May I ask why you ended up there instead of just using the heavy subsidized subscription. I’m actually curious.
eli 3 hours ago [-]
Mimo has subsidized subscriptions too
james2doyle 5 hours ago [-]
2.5 Pro or the regular 2.5?
I always found that those Mimo models to be really good at tool calling and following instructions
esafak 4 hours ago [-]
How fast is it compared with the other Chinese models?
ricardobeat 3 hours ago [-]
They both are in the 50-100 tok/s range. The Mimo v2.5 Pro Ultraspeed beta could reach 1000 tok/s, hoping they can do something similar for the new model, it was amazing.
electroglyph 3 hours ago [-]
[flagged]
NuclearPM 2 hours ago [-]
Real?
yeeeloit 5 hours ago [-]
[flagged]
senordevnyc 5 hours ago [-]
Yeah, this Brazilian dude who has been a contributor here on HN longer than your anonymous account is shilling for a Chinese model company. Makes sense.
platinumrad 5 hours ago [-]
Are you accusing them of astroturfing? Why is it strange for someone to say something topical?
rao-v 22 minutes ago [-]
I absolutely love that someone is doing this! Why isn’t IBM for Granite or Google for Gemini?
If you are going to develop a near frontier model, and you don’t think you have special sauce up your sleeve, why not making training runs and RL environment scores etc. visible to the world?
I’m genuinely learning quite a bit just from the dashboard
dr_dshiv 4 hours ago [-]
Well, if open source AI is dangerous (for OpenAI/Anthropic IPOs?), this is like watching a time bomb.
For my own usage, Luna is cheap enough that I don't care if other models are cheaper. I'm interested if another model is in some way better and not too expensive.
rapind 54 minutes ago [-]
Luna is great but makes a lot of mistakes at high and lower in my experience (large rust codebase). I use Luna Max for asynchronous subagent reviews and am very happy with its work, but it’s slow af.
dzonga 2 hours ago [-]
the open burial started when zAI served their latest model on all Chinese chips.
now we r just noticing the grave getting dug deeper.
ricardobeat 3 hours ago [-]
For reference, Mimo-v2.5-Pro scored 19% on DeepSWE 1.1. This is looking great.
Fable scores 70%, Kimi K3 69%, Astra 74% (all on max effort).
2.6-pro just reached 63.7% by step 10, it's on step 11 right now.
Even flash reached 60.7% by step 12, and it's on step 16 now.
This is so exciting lmao.
passive 4 hours ago [-]
Neat! I've been trying out their next model for the last week, which I assume is a version of this, and it's been a good experience so far.
I had used 2.5-pro for a hefty chunk of development, and found it to work like a somewhat forgetful senior engineer who was new to my project. Very capable, would almost always choose a reasonable option, if not always the best one for the project, and not great at multi-tasking. Generally, made me comfortable not scrutinizing the code line-by-line, but still needed a bit of steering once projects got to a reasonable size.
The next model is a clear step up in the multi-tasking capability at least, with me very rarely having to steer the implementation of a well-defined issue. In terms of code, I found MiMo-V.2.5-pro to be extremely conservative, implementing minimal solutions. The next model seems a little bit more ambitious, in positive ways, making good guesses about gaps/next steps. It also seems to be a fair bit better at design, at least for the little bit I've done, it was good at translating my concepts to practical elements on screen, and cleaned things up nicely as I made suggestions.
krm01 6 hours ago [-]
This is pretty neat. What would be a good reason for the other Model providers to not do this?
kibae 5 hours ago [-]
Speculating here, but I assume researchers can make a reasonable estimate of the size of closed models based on factors like training time, training speed, and the number of tokens processed.
Also, Anthropic and OpenAI probably want to keep each other on their toes so they don’t end up on the wrong side of another Opus 4.6 / GPT-5.3-Codex situation, where one lab releases a model only for the other to drop a better one hours later.
jwpapi 5 hours ago [-]
I think first of all it’s not an obvious idea, also the marketing surplus for other providers is not as big for openai/anthropic as for xiaomi and last but not least I’m pretty sure you can withdraw methodology from here.
I’m saying who has a million dollars for me, so I can make my own model?
fzysingularity 4 hours ago [-]
Very cool to see the openness here, and likely more like this will come from smaller startups where they win users on transparency.
liuliu 6 hours ago [-]
When you run benchmarks while training, isn't that the definition of contamination? Asking because I am not sure if this is normal in big labs now.
jampekka 5 hours ago [-]
Kinda yes. The benchmarks become part of the validation set, which means the models get slightly overfit to them if they are used as criteria for stopping the training. But a lot less compared to using them in the training data.
I'd guess everybody uses at least some benchmarks as stopping criteria, which is kinda sensible, but it also does induce some benchmaxxing, and explains partly why the newest models always tend to eke out in benchmarks.
Correct. If just stopping criteria, that is less contaminated. The question gets muddier once you also use it to determine hyperparameters during small-scale runs.
lucrbvi 5 hours ago [-]
They are using it to evaluate checkpoints during the training, they are probably not using the benchmarks for training the models. It's a common practice for big reinforcement learning runs.
nodja 4 hours ago [-]
They exist to detect degradation. Datasets are not perfect and if a batch contains too much bad data it can ruin a run, also an opportunity to find bad data and improve the dataset filtering.
SwellJoe 5 hours ago [-]
You gotta have something to aim at. And, presumably, the benchmark is not part of the training data, it is the test against which the model is tested at each stage; is behavior moving in the right direction?
esafak 4 hours ago [-]
Not if you don't train against them.
kingstnap 4 hours ago [-]
It's implicitly trained against. There is like information leakage with researchers messing with the training parameters and checkpoints used.
It's not the direct feedback loop of RL but its not far.
ProfessorLayton 5 hours ago [-]
2.6 Pro: >started 2026-09-15 10:32 UTC
For some reason I thought training took much, much longer than what the progress bar suggests.
This is really neat, I'm currently using mimo 2.5 pro, and it's decent (or great given the price). Hopefully their next one is multimodal.
GaggiX 5 hours ago [-]
These are post-training reinforcement learning steps.
krackers 5 hours ago [-]
Yes, updated the submission title to say "post-training" to hopefully prevent further confusion
ttul 49 minutes ago [-]
$5 per second if my eyes don’t fool me. That’s ~$432K per day. Enough to rent 3,000 B300 nodes on Modal.
dr_kiszonka 1 hours ago [-]
Very curious that everyone here (so far) seems to assume this dashboard presents real data.
speedgoose 6 hours ago [-]
I didn't know 2 thirds of the training data would be source code.
jerrygenser 5 hours ago [-]
that is the the "data used to improve the model" when signing up for the subscription plans
leothetechguy 5 hours ago [-]
this is the rl run, not the pretraining run
ahmadyan 5 hours ago [-]
even in pre-training, usually 30%-50% is code these days.
thehamkercat 6 hours ago [-]
This is crazy, but sadly anthropic/openai will never do this, what has happened to this world, where chinese companies are more open than US or even EU companies
medlazik 5 hours ago [-]
Neoliberalism, that famously open and transparent economic ideology
ernsheong 3 hours ago [-]
Mino 2.5 has been my workhorse for coder and tester agents (the ones planner agents delegate tasks to)
rozab 6 hours ago [-]
Why are they doing this? To try head off accusations about distillation?
bayindirh 5 hours ago [-]
Sometimes you're confident about what you're doing and show how you work to the world.
Keeping the garage door open, or at least making the door translucent. It's always cool.
jampekka 5 hours ago [-]
That China's official policy is now to prefer open models and open model development may be a part of it.
Aboutplants 5 hours ago [-]
With that policy in place, labs might be incentivized to be creative in their openness. This being fun/free PR
culi 5 hours ago [-]
BRICS just had a New Delhi meeting where Xi pushed a 5-point plan on AI cooperation that centered on open source models
anemic 4 hours ago [-]
Bottom of the page says "Open is what we value."
wolttam 6 hours ago [-]
Hah, it would be great to see more labs pick this up.
impulser_ 5 hours ago [-]
The Chinese labs are just making fun of the US labs at this point.
Where is the cool shit from the US labs?
culi 5 hours ago [-]
With other software, devs convince their managers of the importance of using open source stuff in their stack. With AI, it's usually managers choosing what models to use for the devs. The US labs don't need to give a damn how much devs like open source
impulser_ 4 hours ago [-]
This isn't about liking open source. This is about the labs just being cool and doing cool shit instead of the opposite which is Anthropic where all they talking about is killing everyone and taking everyone's job.
dlisboa 57 minutes ago [-]
These labs are still (for the time being) made of people, who reflect their lives onto the work.
The US population is much more pessimistic and doomsday driven these days, whereas the Chinese are more optimistic and future driven.
noir_lord 4 hours ago [-]
> The US labs don't need to give a damn how much devs like open source
In the short term, true.
In the long term, unknown but typically when you hold progress that way while other countries don't you at best end up becoming siloed while the rest of the world continues on without you.
That "training cost" is just live revenue count for Anthropic/OpenAI API calls!
/s
heronbank 47 minutes ago [-]
[dead]
Toslink 3 hours ago [-]
[dead]
levocardia 5 hours ago [-]
You'd think they would make it less obvious that they are running their whole operation with Claude
ricardobeat 3 hours ago [-]
If you're thinking of the UI style, definitely not Claude. It is incapable of writing a clear sentence like "what each step's samples are made of", would have used all-caps for everything, more padding and gradients.
jambutters 2 hours ago [-]
They'd be running in the red then cause they charge way less than Claude. Sorry but it just doesn't make logical sense. They have open source, papers, and self hosting too
The cost is unbelievably low, and the quality of intelligence I get is equivalent to when I was working mostly with Anthropic models (late last year/early this year). I'm fully invested in MiMo and I'm very happy with it.
-- PS: I also check almost daily to see if other models are capable of doing such great work. And they do – DS4F is powerful and DS41 is impressive, GLM 5.3 Flash gets a job done well, etc. – but when I add cost of M-token in the ROI math, Jeez! MiMo is an order of magnitude better.
I’ll give Mimo a try.
UltraSpeed was absolutely awesome. I miss it.
DS 4.1 Flash is amazing. Well worth the extra cost.
I always found that those Mimo models to be really good at tool calling and following instructions
If you are going to develop a near frontier model, and you don’t think you have special sauce up your sleeve, why not making training runs and RL environment scores etc. visible to the world?
I’m genuinely learning quite a bit just from the dashboard
https://www.debtdefaultclock.us/
now we r just noticing the grave getting dug deeper.
Fable scores 70%, Kimi K3 69%, Astra 74% (all on max effort).
https://deepswe.datacurve.ai/blog/deepswe-v1-1
Even flash reached 60.7% by step 12, and it's on step 16 now.
This is so exciting lmao.
I had used 2.5-pro for a hefty chunk of development, and found it to work like a somewhat forgetful senior engineer who was new to my project. Very capable, would almost always choose a reasonable option, if not always the best one for the project, and not great at multi-tasking. Generally, made me comfortable not scrutinizing the code line-by-line, but still needed a bit of steering once projects got to a reasonable size.
The next model is a clear step up in the multi-tasking capability at least, with me very rarely having to steer the implementation of a well-defined issue. In terms of code, I found MiMo-V.2.5-pro to be extremely conservative, implementing minimal solutions. The next model seems a little bit more ambitious, in positive ways, making good guesses about gaps/next steps. It also seems to be a fair bit better at design, at least for the little bit I've done, it was good at translating my concepts to practical elements on screen, and cleaned things up nicely as I made suggestions.
Also, Anthropic and OpenAI probably want to keep each other on their toes so they don’t end up on the wrong side of another Opus 4.6 / GPT-5.3-Codex situation, where one lab releases a model only for the other to drop a better one hours later.
I’m saying who has a million dollars for me, so I can make my own model?
I'd guess everybody uses at least some benchmarks as stopping criteria, which is kinda sensible, but it also does induce some benchmaxxing, and explains partly why the newest models always tend to eke out in benchmarks.
https://en.wikipedia.org/wiki/Training,_validation,_and_test...
It's not the direct feedback loop of RL but its not far.
For some reason I thought training took much, much longer than what the progress bar suggests.
This is really neat, I'm currently using mimo 2.5 pro, and it's decent (or great given the price). Hopefully their next one is multimodal.
Keeping the garage door open, or at least making the door translucent. It's always cool.
Where is the cool shit from the US labs?
The US population is much more pessimistic and doomsday driven these days, whereas the Chinese are more optimistic and future driven.
In the short term, true.
In the long term, unknown but typically when you hold progress that way while other countries don't you at best end up becoming siloed while the rest of the world continues on without you.
/s