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

Codestral Mamba

mistral.ai

Post reply on HN
Print view View original

121–130 of 148 posts

Re: Codestral Mamba

#121

Post by PufPufPuf »

karolist wrote: post #119
heeton wrote: post #116

Earlier quoted context omitted.

Have you tried supermaven? It replaced copilot for me a couple of months ago.

I tried it, uses GPT-4o, the $10 sign up credit dissapeared in a few hours of intense coding, I'm not paying $500/mo for a fancy autocomple. Manual instruct style chat about code with Claude-Sonnet-3.5 is the best price/perf I've tried so far, through poe.com I use around 30k credits per day of coding of the 1M monthly allotment, I think it was $200/y. It's not available directly in my country. I've tried a bunch of…

On the pricing page it says $10/month, what you describe sounds like pay-as-you-go, are you sure it was this service you tried?

Re: Codestral Mamba

#122

Post by PufPufPuf »

bhouston wrote: post #8

What are the steps required to get this running in VS Code? If they had linked to the instructions in their post (or better yet a link to a one click install of a VS Code Extension), it would help a lot with adoption. (BTW I consider it malpractice that they are at the top of hacker news with a model that is of great interest to a large portion of the users where and they do not have a monetizable call to action on t…

Currently the best (most user-friendly) way to run models locally is to use Ollama with Continue.dev. This one is not available yet, though: https://github.com/ggerganov/llama.cpp/issues/8519

Re: Codestral Mamba

#123

Post by karolist »

PufPufPuf wrote: post #121
karolist wrote: post #119

Earlier quoted context omitted.

I tried it, uses GPT-4o, the $10 sign up credit dissapeared in a few hours of intense coding, I'm not paying $500/mo for a fancy autocomple. Manual instruct style chat about code with Claude-Sonnet-3.5 is the best price/perf I've tried so far, through poe.com I use around 30k credits per day of coding of the 1M monthly allotment, I think it was $200/y. It's not available directly in my country. I've tried a bunch of…

On the pricing page it says $10/month, what you describe sounds like pay-as-you-go, are you sure it was this service you tried?

Yes that the same, in the "Your Account" section of the service you have "Chat Credits" field, it was seeded with $10 initially and I got it down to $0.11 in one evening. You are right about the pricing page, it does say it's $10/month, but I believe it's $10/month up to some amount of queries in actuality, because they obviously can't offer infinite inference for $10. Maybe I was "holding it wrong", sending too much complete files as part of prompt and the token use and context window skyrocketed, but I absolutely used my initial $10 in one evening, unlike with Claude...

Re: Codestral Mamba

#124

Post by hobofan »

PufPufPuf wrote: post #121
karolist wrote: post #119

Earlier quoted context omitted.

I tried it, uses GPT-4o, the $10 sign up credit dissapeared in a few hours of intense coding, I'm not paying $500/mo for a fancy autocomple. Manual instruct style chat about code with Claude-Sonnet-3.5 is the best price/perf I've tried so far, through poe.com I use around 30k credits per day of coding of the 1M monthly allotment, I think it was $200/y. It's not available directly in my country. I've tried a bunch of…

On the pricing page it says $10/month, what you describe sounds like pay-as-you-go, are you sure it was this service you tried?

From what I can tell inside the product, the public pricing seems deceptive.

It appears that for the $10/month you just get access to additional features (e.g. bigger context) + a budget of $5/month of credits. The credits possibly translate 1:1 to the usage costs of the underlying models you chose.

I asked it to add some missing documentation with the Claude 3.5 Sonnet model to a medium-sized Python file (2k lines) as a first test and that used up 13 cents of the credits. If it were working in a really productive way (where I'd also include more files as context), I'd probably burn through $20-50/day.

I'm wondering why they are not publicly showing some more transparent pricing. Yeah, this method will bump their signup numbers for investors, but it also absolutely wrecks their churn numbers, when people immediately cancel in the first hour.

Re: Codestral Mamba

#125

Post by baq »

treyd wrote: post #92
bhouston wrote: post #46

Earlier quoted context omitted.

On top of Hacker News (the target demographic for coders) without an effective monetizable call to action? What a missed opportunity. Github Copilot makes +100M/year, if not way way more. Having a VS Code extension for Mistral would be a revenue stream if it was one-click and better or cheaper than Github Copilot. It is malpractice in my mind to not be doing this if you are investing in creating coding models.

How the hell does Copilot make $100M/yr? That seems an order of magnitude higher than I would expect at the high end.

yeah exactly only $100M/yr? barely covers expenses

Re: Codestral Mamba

#126

Post by michaelt »

NotMichaelBay wrote: post #91
passion__desire wrote: post #48

Earlier quoted context omitted.

But they also signal competence in the space which means M&A. Or big nation states in future would hire them to produce country models once the space matures as was Emad's vision.

What does a "country model" mean? Optimized for that country's specific language, or with state propaganda or something else?

If you believe LLMs are going to end up built into everything and doing everything, from moderating social media to writing novels and history books, making such a model will be the most political thing that has ever happened.

If your country believes guns=bad nipples=good war=hell but you get your novels and history books written by an LLM trained by people who believe guns=good nipples=bad war=heroic it would be naive to expect the output to reflect your values and not theirs.

Even close allies of the US would be nervous to have such power in the hands of American multinational corporations alone - so the French state could be very eager for Mistral to produce a competitive product.

Re: Codestral Mamba

#127

Post by whywhywhywhy »

mountainriver wrote: post #56
rjurney wrote: post #50

Earlier quoted context omitted.

Claude is much better. Overwhelmingly better. It not only implements deep learning models for me, it has great suggestions on evolving them to actually work.

lol no it’s not, the benchmarks don’t show that at all. Both have issues in different ways

Benchmarks claim GPT4o is better than GPT4 so anyone who's actually used the software knows benchmarks mean nothing.

Re: Codestral Mamba

#129

Post by jessicant »

PufPufPuf wrote: post #121
karolist wrote: post #119

Earlier quoted context omitted.

I tried it, uses GPT-4o, the $10 sign up credit dissapeared in a few hours of intense coding, I'm not paying $500/mo for a fancy autocomple. Manual instruct style chat about code with Claude-Sonnet-3.5 is the best price/perf I've tried so far, through poe.com I use around 30k credits per day of coding of the 1M monthly allotment, I think it was $200/y. It's not available directly in my country. I've tried a bunch of…

On the pricing page it says $10/month, what you describe sounds like pay-as-you-go, are you sure it was this service you tried?

Supermaven has a $10/month subscription which covers unlimited use of the in editor code autocomplete stuff.

It also has a pay as you go chat which is integrated into the IDE w/ hotkeys, however the above subscription gives you $5 free credits for it a month. This doesn't call their own model but stuff like gpt-4o which is presumably why they don't offer unlimited free usage.

Re: Codestral Mamba

#130

Post by DalasNoin »

bhouston wrote: post #8

What are the steps required to get this running in VS Code? If they had linked to the instructions in their post (or better yet a link to a one click install of a VS Code Extension), it would help a lot with adoption. (BTW I consider it malpractice that they are at the top of hacker news with a model that is of great interest to a large portion of the users where and they do not have a monetizable call to action on t…

I feel like local models could be an amazing coding experience because you could disconnect from the internet. Usually I need to open chatgpt or google every so often to solve some issue or generate some function, but this also introduces so many distractions. imagine being able to turn off internet completely and only have a chat assistant that runs locally. I fear though that it is just going to be a bit to slow at generating tokens on CPU to not be annoying.
Post reply on HN