Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

mimo.xiaomi.com

Post reply on HN
Print view View original

231–240 of 512 posts

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#231

Post by visarga »

serpix wrote: post #16

I may sound like a shill, but exponential growth and all. We are going to get near instant software from prompt, multiple ones and then choose the best one. Discussions about choosing a library with the best syntactic sugar method naming is just as crazy as suggesting we type in assembly.

> We are going to get near instant software from prompt, multiple ones and then choose the best one.

If you extract the spec from first implementation and reimplement from scratch you get a free testing oracle. Where they diverge you send the agent to decide which one had a bug.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#232

Post by pixel_popping »

AlecSchueler wrote: post #226
binyu wrote: post #220

Earlier quoted context omitted.

> Right now Claude is faster than me on some tasks but we’re at least close. I dont doubt it, but I don't think you can spawn 10 copies of yourself working simultaneously.

No, but nor can you keep track of what 10 agents are doing simultaneously. Hence the multitasking regret.

An agent can, you don't need to watch tasks, you can have a live digest with another tool.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#233

Post by himata4113 »

vanuatu wrote: post #212
himata4113 wrote: post #190

Earlier quoted context omitted.

I was saying that AI is going to make software development cheaper as in the salaries of software engineers will go down because some of that salary will now be redirected to AI companies and the fact that the world will need to absorb twice-(x10?) the amount of the development power.

its not obvious to me that salaries go down, my hunch was that salaries go up but the bar is higher. Software becoming easier to produce (still hard to verify and make useful fwiw) raises the ambitions of software projects, and we don't seem to be close to the ceiling of demand for software systems

There's a limit to what the demandXsupply curve can absorb. It really depends if there's twice as many developers or 10 times more. I think we have enough software development jobs to where we can absorb productivity doubling rather easily, not so sure about anything beyond that.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#234

Post by npn »

2001zhaozhao wrote: post #201
npn wrote: post #9

How? edit: now I read the article fully, seems like they utilize some very effective MTP algorithm. and somehow the quality is still decent enough. though, I doubt that the quality really only drip a bit like they claimed. maybe for the benchmarks, but for general uses the heavily quantized models very often so worse result.

i wonder if it will be possible to hardcode a model with some kind of MTP-adjacent algorithm to use a smaller portion of it to generate most of the tokens but route to the real experts every once in a while to steer it towards good thinking directions. (Perhaps this is done only when it's generating its thinking block, and the training takes it into account) Could result in very high efficiency and still good intelli…

I doubt you can do that. MTP magic happens because for texts, we have a lot of low value fixed tokens that almost always get generated in the sequence (like punctuation, function words, language keywords etc). for most important ones (the entities, the content words, variables) you still need the full model.

so there is alwasy a maximum limit for how well MTP can do.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#235

Post by wuliwong »

paulinho1 wrote: post #26
nkmnz wrote: post #15

Earlier quoted context omitted.

No idea why you've been downvoted. This is excellent news.

Because this never gets brought up about US models, which have just as much censorship as the Chinese ones.

You should read OPs responses in this thread. He actually does test US models. ¯\_(ツ)_/¯

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#236

Post by throw1234567891 »

throwaway67678 wrote: post #170
flexagoon wrote: post #92

Earlier quoted context omitted.

I'm using Deepseek-v4-pro as my main model and this is sometimes pretty annoying, I have to do some easy boring task, think "I'll just leave the agent to do it and go take a nap", but it's already done writing the code before I even walk away from the computer

Agent mania setting in It's also pretty funny sometimes how it gives weird future roadmap estimates ("part 2 - 3 weeks, part 3 - 2 months", etc.) and when you tell it to actually do those changes it's pretty much done in half an hour

It repeats what it has seen in the training data. Expecting it to reason about the complexity of a task is a pipe dream. The best is to tell it not to come back with estimates, and when it does, remove them anyway.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#237

Post by vanuatu »

himata4113 wrote: post #233
vanuatu wrote: post #212

Earlier quoted context omitted.

its not obvious to me that salaries go down, my hunch was that salaries go up but the bar is higher. Software becoming easier to produce (still hard to verify and make useful fwiw) raises the ambitions of software projects, and we don't seem to be close to the ceiling of demand for software systems

There's a limit to what the demandXsupply curve can absorb. It really depends if there's twice as many developers or 10 times more. I think we have enough software development jobs to where we can absorb productivity doubling rather easily, not so sure about anything beyond that.

True on the demand/supply curve

I think due to how leveraged software is, the top % of software developers are more desired (and compensated) than ever, and the bottom % will have difficulty finding a role, and there are structural barriers to entering that top % (intelligence, location, etc). Companies have infinite demand for the cream of the crop talent

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#238

Post by bscphil »

adrian_b wrote: post #181
wolttam wrote: post #41

Earlier quoted context omitted.

I'd love to know of such an example where a U.S. LLM blatantly denies something factual. Maybe I'm living under a rock but I can't think of one

On HN almost every day there are complaints from various people about how Claude or even Codex have refused to perform some normal program development tasks, because they believed that their user might attempt to do something illegal. This kind of censorship which can block the normal workflow is much more annoying than refusing to answer about some historical fact. Moreover, even when they are used conversationally…

> about how Claude or even Codex have refused to perform some normal program development tasks

> a lot of reports that the US LLMs refuse to answer questions

I think the specific ask is for a case where the LLM is trained to lie about something. What you've come up with are cases where it refuses to do something, possibly for legal reasons but maybe not (you can come up with plausible non-legal reasons why a company training an LLM might want it to refuse to give you instructions on making a bomb, even if instructions on making a bomb are protected First Amendment speech).

An LLM that responds with "I'm sorry, due to legal requirements placed on my creators, I'm unable to answer questions about events at Tiananmen square in 1989." strikes me as much less problematic than one that pretends there is no relevant or reliable information that exists, or explicitly supports a regime narrative. But I'm also of the opinion that an LLM refusing to help you build a fertilizer bomb is much more reasonable than one that suppresses information of a political nature. I can't think of a case where information that reflects the broad consensus of experts is suppressed by US based LLMs for political reasons.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#239

Post by polski-g »

petesergeant wrote: post #189
ilaksh wrote: post #180

Earlier quoted context omitted.

I'm kind of poor so I have been trying to use DeepSeek v4 Flash, GLM 5.1 etc. as much as possible recently instead of Claude or GPT.

You would do us all a service by telling us how your experiences of that have been.

I used Opus 4.6, then downgraded to Sonnet, then to GLM5/5.1. GLM is as good as Sonnet. I recently started using Opus 4.8 again and GLM is not close to that.

30 day eval for each.

Re: MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

#240

Post by yunohn »

goyozi wrote: post #130
ipkstef wrote: post #65

Earlier quoted context omitted.

asking for curiosities sake. What kind of PR loop are you running that takes a few hours?

I’m rewriting our integration test suite to run tests in parallel. I have the changes split across 7 branches, and each needs to be fixed to have no flaky tests. I told it I want 3 consecutive CI runs with no flakes and no artificial fixes / assert removals etc. We’ll see what comes out; it’s almost a side project so there’s not much to lose other than some of my weekly limit that resets soon.

> a side project so there’s not much to lose other than some of my weekly limit that resets soon

Basically the entire token-maxxing AI hype train in a nutshell. Lovely!

Post reply on HN