If only they had launched that yesterday I might have avoided Copilot auto model selection using a 9x model, quietly burning my monthly quota in a single afternoon.
MAI-Code-1-Flash
71–80 of 297 posts
Re: MAI-Code-1-Flash
#72Mark Zuckerberg must be in crisis. Microsoft releasing models that compete with Claude's models. Meanwhile the only thing anyone knows about Mark's models is that they help you get hacked more easily.
Wait… I think he has moltbook IP as well that he can scale up.
Seriously tho, wtf is going on over at Meta? Anyone working there currently want to describe the vibe of the org when it comes to being a frontier company?
Re: MAI-Code-1-Flash
#73Does 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.
Claude code itself spins a lot of its subagents with Haiku. The model has low hallucination rate, so it is great for exploration tasks. I guess this is what the best purpose of this model here will be as well. Which is a lot of tokens - many tasks spin multiple exploration agents before the planning or fixing, that is then just a few tool calls.
Re: MAI-Code-1-Flash
#74is 51% good enough to reliably use? There's no world in which I use an AI agent where it gets even 15% of the code wrong, that's as bad a Tesla FSD where you need to pay attention to the road while engaging FSD. What's the point? My attention is what I'm trying to relieve, not mostly correct functionality. The only thing that matters is whether you can one-shot code like Claude or Codex, I'm not interested in a small…
51% does not mean it randomly gets things wrong half the time.
These things can be useful if you can accurately predict which tasks they will reliably do, and which they will usually fail on. Then you can get much more reliable work from them.
Re: MAI-Code-1-Flash
#75Mark Zuckerberg must be in crisis. Microsoft releasing models that compete with Claude's models. Meanwhile the only thing anyone knows about Mark's models is that they help you get hacked more easily.
Wait… I think he has moltbook IP as well that he can scale up. Seriously tho, wtf is going on over at Meta? Anyone working there currently want to describe the vibe of the org when it comes to being a frontier company?
I don't understand his plan, if I were him I'd either have just gone all in on making RAM which would become very lucrative, or would have focused on building programming models. They've built some key open source technologies, but its as if Mark Zuckerberg cannot run anything that isn't a social media company / project.
Re: MAI-Code-1-Flash
#76Earlier quoted context omitted.
If you don't hit a limit running Opus, it means you are very much in the loop. For example you probably don't have days where you ask Opus to review your whole code base and look for code duplication/technical debt/robustness issues, and then to fix some of the found issues, and do this 3-5 times until no big issues are found anymore.
What’s your prompt for this, the way you described it made it seem like there’s a generalizable way I can go about this. I just rely on a testing pipeline instead so can’t think of why I would need to proactively find holes where tests haven’t already done that for me.
tests will not find inconsistent naming, duplicate functions, scenarios you have not thought about testing
I use quite plain prompts, nothing fancy:
> go over the tests and do a code review, focusing on how well they test inventory management, planner and controller. maybe some tests need to be deleted, maybe other tests need to be added. the end goal should be good coverage of the core features.
> do a code review, focusing on robustness/correctness issues. validate that the code correctly implements specification.md. focus on the async client.
> there was a big refactor. please do a code review, focusing on eliminating tech debt. look for unused, obsolete or duplicate code that can be removed, look for mismatched interfaces, inconsistent function/argument/variable names. do not output what is correct, just the issues you found. for each issue output instructions for a coding agent on how to fix it. do not nitpick.
Re: MAI-Code-1-Flash
#77Re: MAI-Code-1-Flash
#78Earlier quoted context omitted.
If you don't hit a limit running Opus, it means you are very much in the loop. For example you probably don't have days where you ask Opus to review your whole code base and look for code duplication/technical debt/robustness issues, and then to fix some of the found issues, and do this 3-5 times until no big issues are found anymore.
What’s your prompt for this, the way you described it made it seem like there’s a generalizable way I can go about this. I just rely on a testing pipeline instead so can’t think of why I would need to proactively find holes where tests haven’t already done that for me.
I use similar workflow. Here is my refactoring and code quality prompt that I regularly run:
Perform a thorough analysis of the project (the code and the documentation).
- Explore the project, go over all important files one by one and look for any mistakes or possible bugs.
- Look for refactoring opportunities and ways to improve code quality and organization.
- Identify any potential cruft/bloat, to ensure our code is clean and logically laid out. Keep in mind that efficient and good quality code needs to avoid over-engineered constructs and needless complexity. Avoid complicated logic where simple solutions would be more elegant.
- Pay attention to comments: There should be enough of them to document the intent and provide high-level overview of the code logic, but not too much; avoid/remove excessive comments that simply restate the code logic or do not provide any useful information.
- Every important function should have a top-level docstring comment that clearly explains its purpose, high-level logic overview, arguments, and return values.
- Analyze the names of constants/variables/functions/classes and other code elements: could some of them be renamed to make their purpose more clear?
- Analyze the documentation, uncover any potential inaccuracies/omissions and ensure the docs reflect the code.
- Brainstorm ideas for improvements of the code and docs.
After you finish the analysis, save an analysis report into "_analysis_report.md" in the project root folder.Re: MAI-Code-1-Flash
#79Does 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.
I actually find planning/design easier with a smaller model and implementation with a larger one. I'm mostly manually working with the model on planning and design and decisions are mine and smaller models are faster. And when there's a clear design/wayforward, the bigger models are usually better at understanding the overall context and applying the specific patch they were assigned to. I call it the 1-2 punch system where you do the first light punch then the harder punch when its actually important to hit properly. I know it goes against the standard of throwing the biggest model at design but I personally experience the bigger models try to do TOO MUCH and take a lot of time which is something that's not good in the design/arch/boilterplate phase.
Re: MAI-Code-1-Flash
#80Does 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.
I keep trying to, because I really want to make qwen 3.6 35b work for end implementation of a fleshed out spec (mostly for local data privacy reasons).
...but I spend so much more time correcting it, or building pipelines to try, retry, and converge, that it's rarely worthwhile for me in either time or $ spent vs Opus.