381–390 of 418 posts
In the end, a business gets to decide how and when to strategically subsidize their users.
Earlier quoted context omitted.
This argument is predicated on Anthropic losing money on the subs, but I'm not sure that's a cut and dried argument. OpenAI have said publicly that they're (very) profitable on inference, and they're much cheaper than Anthropic. I suspect this is just artificially trying to create a moat. The problem is their moat is not as sticky as they think it is - I completely ditched Claude for Codex a while ago, my money now g…
They aren't public companies (yet). They are allowed to just lie about these things. It's also not really reasonable to only count inference compute as a cost since it's not like any of these companies could stop doing R&D without being abandoned for having worse models within a year or 2
That would turn into investment fraud the moment they IPO.
Earlier quoted context omitted.
Or the Python or Typescript Agent SDK libraries: https://platform.claude.com/docs/en/agent-sdk/overview
Or can you? It's my understanding that you cannot use your subscription with the Agent SDK, that's what the docs say: > Unless previously approved, Anthropic does not allow third party developers to offer claude.ai login or rate limits for their products, including agents built on the Claude Agent SDK. Please use the API key authentication methods described in this document instead. Though there was that tweet [0] a…
> Unless previously approved, Anthropic does not allow third party developers to offer claude.ai login or rate limits for their products
So you can't offer a product based on the subscription. I replied to a comment proposing to use the CLI with --prompt:
> you can also use Claude Code programmatically with the `prompt` parameter in the cli.
And I just informed that Agent SDK is a much nicer way to do that. For internal use or whatever the ToS allows.
Since there's a lot of questions about what this means, let me explain. Anthropic has two different products that are relevant here: the Claude API and Claude Code. The Claude API has usage based pricing. The more you use, the more you pay. With Claude Code, you can get a monthly subscription which gives you a fixed amount of usage. Comparing equivalent token generation between the Claude API and Claude Code, Claude…
The part I never really understood, was I thought the subscriptions were to try and boost Opus usage, not claude code usage ? I'm not sure why they care whether you use API or claude, as they limit the number of tokens you can use anyway - and once the request hits the model, I would have thought it takes the same amount of effort to process it regardless of where it comes from ?
Since there's a lot of questions about what this means, let me explain. Anthropic has two different products that are relevant here: the Claude API and Claude Code. The Claude API has usage based pricing. The more you use, the more you pay. With Claude Code, you can get a monthly subscription which gives you a fixed amount of usage. Comparing equivalent token generation between the Claude API and Claude Code, Claude…
But to use their product,you have to go through the non-human-friendly API route, or else it's against the rules and Anthropic will sic their legal team onto you...
Something about this reasoning seems brittle. Specially in a world of Agentic tools
Ok
Earlier quoted context omitted.
That doesn't make sense. The PS5 is subsidized because the make money with the games. Printers are subsidized because the make money with the ink. The API use is subsidized because they make money with Claude Code? I would understand if Claude Code could only be used with Anthropics API but not the other way around. 1 million tokens is 1 million tokens unless Claude Code is burning tokens and others are more efficien…
They want you to become dependent on Claude Code, so that later they can milk you.
As much as it kills the living shit out of me to say this... Apple M5 Ultra will probably mean for a lot of us that we wont need openai or anthropic for coding llms anymore
Local models are coming boys. As much as it kills the living shit out of me to say this... Apple M5 Ultra will probably mean for a lot of us that we wont need openai or anthropic for coding llms anymore
While SOC is definitely the future or local AI, even if you get an M5 that's jacked to the tits, you still won't be able to store the entire model in unified memory on top of the OS and whatever other applications you have running. 128GB is the upper limit for unified memory on the M5, which on paper could support a model like gpt-oss:120b but still with a nerfed context size and quantised at that. Furthermore, the cost of a maxed out Macbook Pro M5 Max is between $8-10k depending on your storage option, screen size etc., so we can safely assume that the M5 Ultra will be even more. There's also no guarantee that the Ultra will offer double the amount of unified memory, it may only offer more cores but cores aren't the current bottleneck for local AI, memory is.
If you consider what you'd be paying above and beyond your requirements barring local AI, it would be adding $5-6k to the price tag at a minimum. That equates to 5 years worth of a Claude Code subscription! Even if you shouldered that cost with your NFT fortunes you likely wouldn't achieve performance parity with CC.
I'm equally as excited as you are about the future for local AI and I am actively working in this space everyday to improve it but we're still a long way off from being able to match model size, context size, token/sec, TTFB etc. A single H100 is so OP, so hosting thousands in a data centre it's expected that it should remain unrivalled.
The area where I'm having success with local AI is by pairing local models with other supportive technologies like databases and the like to compensate for smaller context size. There are still many in roads to be made in this area and that bodes well for the future of local AI as models become more efficient and sophisticated.