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

MAI-Code-1-Flash

microsoft.ai

Post reply on HN
Print view View original

251–260 of 297 posts

Re: MAI-Code-1-Flash

#251

Post by Hfuffzehn »

Hfuffzehn wrote: post #247

So I guess the important link the marketing department forgot is this one: https://docs.github.com/en/copilot/reference/copilot-billing... Model Input Cached input Output MAI-Code-1-Flash $0.75 $0.075 $4.50 Comparing to Claude Haiku 4.5 $1.00 $0.10 $5.00 looks fine. But they also forgot to include the benchmarks comparing to GPT-5.4 mini $0.75 $0.075 $4.50 Those would have been helpful.

And as I am on holiday today I will try to help them out:

                   GPT-5.4 mini Haiku 4.5 MAI-Code
SWE-Bench Pro 54.4 % 35.2% 51.2%

Terminal-Bench 2.0 60.0 % 41.6% 54.8%

Source: https://openai.com/index/introducing-gpt-5-4-mini-and-nano/

Re: MAI-Code-1-Flash

#252

Post by Krysoph »

HarHarVeryFunny wrote: post #250
easygenes wrote: post #198

Earlier quoted context omitted.

While I agree directionally, I'll caveat that "cost per token" != "cost per task". In the case of Qwen3.6 it tends to think 1.6x more than Haiku, so the cost of Haiku on the same tasks tends to only be about double. More detail from comparing their Artificial Analysis metrics: Qwen3.6-35B-A3B vs Claude Haiku 4.5 reasoning mode · AA Intelligence Index v4.0 46.0 ┤ ↖ better — cheaper · smarter · faster │ │ 44.0 ┤ ╭─────…

How did you get that nicely formatted graph and table in your post ?!

> Text after a blank line that is indented by two or more spaces is formatted as code.

https://news.ycombinator.com/formatdoc

  crimes ↑
         │
   10.0  ┤                                           ● Airport burger
         │                                      ╭──────────────╮
    8.0  ┤                                      │  theft arc   │
         │                                      ╰──────────────╯
    6.0  ┤                         ● Five Guys
         │
    4.0  ┤              ● Food truck burger
         │
    2.0  ┤      ● McBurger
         │
    0.0  ┤ ● Homemade burger
         │
         └───────┬─────────┬─────────┬─────────┬─────────→ price
                $2        $8        $14       $22       $38

  ┌────────────────────┬────────┬──────────────┬────────────────────┐
  │ burger             │ price  │ crime index  │ expected behavior  │
  ├────────────────────┼────────┼──────────────┼────────────────────┤
  │ Homemade burger    │   $2   │          0.0 │ law-abiding citizen│
  │ McBurger           │   $6   │          1.4 │ steals extra napkin│
  │ Food truck burger  │  $11   │          3.1 │ lies about hunger  │
  │ Five Guys          │  $18   │          6.2 │ financial crime    │
  │ Airport burger     │  $34   │          9.7 │ enters villain arc │
  └────────────────────┴────────┴──────────────┴────────────────────┘

  conclusion: burger inflation is a gateway condiment

Re: MAI-Code-1-Flash

#253

Post by HarHarVeryFunny »

dist-epoch wrote: post #59
redrove wrote: post #27

Earlier quoted context omitted.

It’s about bang for buck. That high a score for 5B params is pretty good, nigh unbelievable a short while ago. It is my belief that smaller models will get better and better, and even cloud SOTA models will shrink. Yet another reason the current buildout will feel like the railroads.

The SOTA models will not shrink, because the problems will get bigger, from "write me a C compiler" to "clone Stripe business and run it".

There will always be tasks that are withing reach of whatever the SOTA models are, but not of the cheaper, perhaps locally runnable ones. It seems that already people are finding Qwen 3.6 27B sufficient for many coding tasks (the llama.cpp author is now using it exclusively).

As models get better and smaller, I expect that we will rapidly (within a year?) get to the point where SOTA models are not needed for the vast majority of coding tasks, and even today it seems many people are just using them for the planning phase.

How many people drive Ferraris vs Fords? How many people driving a Ford would, on a utilitarian basis, be any better off driving a Ferrari?

So far there seems to be mainly two high volume use cases that have been found for LLMs - coding and business flow automation, and it seems neither of these need SOTA models. I wonder if there will continue to be enough market demand for massive expensive SOTA models to make them worthwhile developing?

Re: MAI-Code-1-Flash

#254

Post by HarHarVeryFunny »

adrian_b wrote: post #221
tosh wrote: post #15

not open weight or at least I did not find anything indicating open weight

Tomorrow NVIDIA will publish Nemotron 3 Ultra, which will be the biggest open weights LLM from a US company (550B parameters). The early testers have confirmed that it is much better than all earlier US open weights models, but it is not as good as the best Chinese open weights models. While Nemotron 3 Ultra is not the smartest open weights LLM, it is well optimized for fast inference, so it is much faster than the o…

NVIDIA seem to be following a smart Intel-like strategy of selling chips and also creating software that helps create demand for those chips. With Intel it was things like MKL, IPP, OpenCV etc, and with NVIDIA it is not just CUDA and development libraries but also models like Nemotron.

The pure-AI companies like OpenAI and Anthropic are hoping to sell you API access to cloud-based AI, perhaps running on NVIDIA chips, but it seems NVIDIA's plan may be for you to run local AI, maybe from NVIDIA, running on local NVIDIA chips.

Re: MAI-Code-1-Flash

#255

Post by fmajid »

yorwba wrote: post #227
stingraycharles wrote: post #131

Earlier quoted context omitted.

I understand what you’re saying, but I am generally very careful when comparing models and their benchmarks; benchmarks often don’t really match “real world” quality.

The technical report https://microsoft.ai/wp-content/uploads/2026/06/main_2026060... has a lot of detail about decontaminating their training data and developing new in-house benchmarks to ensure reliable evaluation. If other models were just overfit to public benchmarks while Microsoft produced something that generalizes better to unseen data, they could've used those in-house benchmarks to argue that point. Instead…

If their model was trained purely on properly licensed data, the reduced legal liability could be a selling point

Re: MAI-Code-1-Flash

#256

Post by HarHarVeryFunny »

Krysoph wrote: post #252
HarHarVeryFunny wrote: post #250

Earlier quoted context omitted.

How did you get that nicely formatted graph and table in your post ?!

> Text after a blank line that is indented by two or more spaces is formatted as code. https://news.ycombinator.com/formatdoc crimes ↑ │ 10.0 ┤ ● Airport burger │ ╭──────────────╮ 8.0 ┤ │ theft arc │ │ ╰──────────────╯ 6.0 ┤ ● Five Guys │ 4.0 ┤ ● Food truck burger │ 2.0 ┤ ● McBurger │ 0.0 ┤ ● Homemade burger │ └───────┬─────────┬─────────┬─────────┬─────────→ price $2 $8 $14 $22 $38 ┌────────────────────┬────────┬───…

Thanks, so in this case the value of "code fomatting" is using a fixed-width font ?

The next question is where did the "ASCII-art" graph and table come from? Are there sites to generate these?

Re: MAI-Code-1-Flash

#257

Post by Aperocky »

bel8 wrote: post #82

It's a start and I welcome competition but I don't think I ever used small cloud models like Haiku 4.5. They are cute but for serious coding they tend to waste your expensive time. And this certainly wont bring me back to GitHub Copilot which I cancelled yesterday. GitHub Copilot had competitive pricing until yesterday when they changed from per-request to one of the most expensive per-token quotas. Seriously, take a…

If you use claude-code Haiku is used under the hood for certain task. I'm not sure what it is, but there's some kind of routing that goes to Haiku automatically.

Re: MAI-Code-1-Flash

#258

Post by wd021 »

hmokiguess wrote: post #33

Does anyone actually uses these smaller models for coding? If so, how? I usually Opus everything. Is the play to plan/design/architect with a heavier model than delegate structured tasks to these smaller ones? Would appreciate to hear someone's opinion on having done and tested both paths.

[dead]

Re: MAI-Code-1-Flash

#259

Post by yowlingcat »

papascrubs wrote: post #177
GaryBluto wrote: post #174

Earlier quoted context omitted.

Using OpenRouter, going to migrate to DeepSeek's official API soon. I'm not using it for anything commercial or for private data so I have no privacy qualms.

Makes sense. Privacy is my only real hang up with DeepSeek. Both of the big SOTA providers have become extremely filtered. Things that I could do one version ago are now getting refusals. Anthropic is almost unusable. ChatGPT is slightly better. Even with a "cyber exception" in place and a vetted account. They are going to force me to take my business elsewhere.

You can always run deepseek yourself, v4-pro and flash are open weights. It's a little tricky to get the hang of self deploying open weight models but you do fully own your deployment substrate and privacy narrative at that point.
Post reply on HN