I use it for my own work, but I think the reputation is currently overstated. The big three each have fairly similar performance when using their best models.
I use it for my own work, but I think the reputation is currently overstated. The big three each have fairly similar performance when using their best models.
Hey, I wrote that! For the reference I didn't mean that to say that Claude is completely untrustworthy, but just like any AI you should actually check it's reasoning and ask it to find sources for...
Hey, I wrote that!
For the reference I didn't mean that to say that Claude is completely untrustworthy, but just like any AI you should actually check it's reasoning and ask it to find sources for its information. Claude is less sycophantic than other mega-sized LLMs that I've used but it's still there. When I tried asking it about local communities and we went back and forth on if a particular group was part of it - it said it was, I thought it wasn't and it agreed with me, then I did more research on it and it turns out that it actually was and Claude agreed with me again. You basically have to use it as a research assistant rather than a full-blown researcher.
As far as I can tell (and Claude backs me up on this), Anthropic doesn't have any business dealings with xAI or any of Musk's companies; it just brings up Grokipedia articles because it was the one that it came up with that seemed to fit the best. Heck, it might be getting picked because the articles are long-winded and have the appearance of authority.
Fair enough. I wonder if it’s possible to get Claude to exclude certain sources? I suppose I could check but… I won’t. I know me, I’ll forget by the time I close this tab.
Fair enough. I wonder if it’s possible to get Claude to exclude certain sources? I suppose I could check but… I won’t. I know me, I’ll forget by the time I close this tab.
Your profile has a “personal preferences” section which is global to all conversations. With Claude Code you can make an installation level or project level preferences file named CLAUDE.md. I...
Your profile has a “personal preferences” section which is global to all conversations. With Claude Code you can make an installation level or project level preferences file named CLAUDE.md. I think there’s something simelar for Claude Desktop, but that’s a step too far for me to try long enough to find out.
Edit: also forgot you can use Projects for the phone and web apps, which let you add in context in a similar manner, including using other types of documents like PDFs and images.
I also immediately canceled and removed OpenAI after the pentagon nonsense. I switched to Anthropic and I am pushing my company to do the same. I think Altman miscalculated by grabbing that...
I also immediately canceled and removed OpenAI after the pentagon nonsense. I switched to Anthropic and I am pushing my company to do the same.
I think Altman miscalculated by grabbing that opportunity. AI is kind of becoming a commodity. The difference between each companies flagship models isn’t huge.
As models plateau I suspect it’s going to become more about orchestration and efficiency. IE having one model use other more efficient models as subagents for specific components of your request.
As models plateau I suspect it’s going to become more about orchestration and efficiency.
IE having one model use other more efficient models as subagents for specific components of your request.
I did the exact same thing. I tested Gemini a little, first, but I'm disturbed by the amount of data I've given to Alphabet over the years, and I'd like to slow that bleed down. Sticking to their...
I did the exact same thing. I tested Gemini a little, first, but I'm disturbed by the amount of data I've given to Alphabet over the years, and I'd like to slow that bleed down.
Sticking to their values definately gave them good will with consumers. From a corporate standpoint, IT is more likely to build partnerships with the companies they use at home.
I mentioned this situation to my wife recently (she's too busy to watch all the tech drama in the world) and I'm pretty sure she swapped to Claude or will swap to it. She's using it in her...
I mentioned this situation to my wife recently (she's too busy to watch all the tech drama in the world) and I'm pretty sure she swapped to Claude or will swap to it. She's using it in her business and when we talked about it I began to really realize that for her workload there is absolutely nothing tying her to one service; she deletes one app and installs another. Maybe Claude is better for coding and Grok is better at being a cringe edgelord, but these companies may have to start worrying about their reputation more.
Edit: She did switch to Claude and says its better in some ways but worse in others. Overall I think she's happy.
Honest question to anyone who knows these things: is Grok actually meaningfully better at any given metric than any other AI? I feel that it’s only ever mentioned because it’s going amok on X...
Honest question to anyone who knows these things: is Grok actually meaningfully better at any given metric than any other AI? I feel that it’s only ever mentioned because it’s going amok on X (which I am amazed is still culturally relevant, itself).
I also got rid of ChatGPT and settled on Claude exclusively after the Pentagon thing. But I doubt individuals move the needle on their revenue when I've heard of companies giving devs >$1k per...
I also got rid of ChatGPT and settled on Claude exclusively after the Pentagon thing. But I doubt individuals move the needle on their revenue when I've heard of companies giving devs >$1k per month budgets for Claude.
Well, some of us, having chosen a favourite for personal work, choose that favourite among the options at their workplace (where they spend that >$1k per month budget).
Well, some of us, having chosen a favourite for personal work, choose that favourite among the options at their workplace (where they spend that >$1k per month budget).
Anthropic also said Monday that its revenue run rate has now crossed $30 billion on an annualized basis — more than three times the roughly $9 billion figure it recorded at the end of 2025. Enterprise traction has also accelerated: The number of clients committing at least $1 million a year has surpassed 1,000, a threshold Anthropic said is twice what it was reporting around the time of its Series G announcement in February.
[...]
Anthropic said the majority of the new infrastructure will be built on U.S. soil, framing the commitment as a continuation of a pledge made last year to direct $50 billion toward domestic computing capacity. According to Monday's securities filing, Broadcom flagged that Anthropic's ability to draw on the additional compute hinges on its ongoing commercial performance, and noted that discussions with outside operational and financial partners are underway to support the rollout.
It's also a garbage chart of annualized revenue. I'll be impressed when the show actual revenue next to expenses and not accounting magic tricks intended to hype gullible investors. Hypothetical...
It's also a garbage chart of annualized revenue.
I'll be impressed when the show actual revenue next to expenses and not accounting magic tricks intended to hype gullible investors.
Hypothetical example:
I have an annualized revenue of $1 million. All I had to do was to project out my earnings this month which coincidentally included my parents dying and leaving me $10,000.
Another way to put this is:
In March, we earned an extra $2,500,000. If that's because a bunch of businesses bought new annual contracts, or price hikes, that's 1/12th the reported number.
This isn't an issue of annualized revenue, its an issue of accounting fraud where you're booking one-time sales as recurring revenue. This situation is a combination of accounting fraud and Annual...
All I had to do was to project out my earnings this month which coincidentally included my parents dying and leaving me $10,000
This isn't an issue of annualized revenue, its an issue of accounting fraud where you're booking one-time sales as recurring revenue.
In March, we earned an extra $2,500,000. If that's because a bunch of businesses bought new annual contracts, or price hikes, that's 1/12th the reported number.
This situation is a combination of accounting fraud and Annual Contract Value here, not annualized revenue.
Source: I just read the Enron book, where the author goes through a variety of accounting frauds in great detail.
It definitely sounds like they are, or, at the very least, not being very honest in the way they're reporting. E: Also, while perhaps not fraud, they said they calculate their annualized revenue...
It definitely sounds like they are, or, at the very least, not being very honest in the way they're reporting.
Non-GAAP gross margins [are] calculated by Anthropic management…[are] unaudited, company-provided, and may not be comparable to other companies.”
E: Also, while perhaps not fraud, they said they calculate their annualized revenue by multiplying the last 4 weeks revenue times 13... and then report it right after the launch their fancy new models, a time when it's very likely that people are trying them out and then some will stay but many will cancel their subscriptions.
Yes, they're company-provided numbers that shouldn't be trusted as much as audited financial statements. But this isn't itself evidence that they're lying.
Yes, they're company-provided numbers that shouldn't be trusted as much as audited financial statements. But this isn't itself evidence that they're lying.
I wasn't totally clear, but I agree with you. I don't have a reason to think Anthropic is committing accounting fraud. I was trying to say that the listed examples were of clear fraud, not...
I wasn't totally clear, but I agree with you. I don't have a reason to think Anthropic is committing accounting fraud. I was trying to say that the listed examples were of clear fraud, not legitimate caveats of annualized revenue numbers.
Annualized revenue does not mean they earned that much (since it hasn't been a year), so the big numbers don't quite mean what they seem. We are also missing the financial statements that would be...
Annualized revenue does not mean they earned that much (since it hasn't been a year), so the big numbers don't quite mean what they seem. We are also missing the financial statements that would be available for a public company, so we have to go by these selective disclosures. Someone put them in a graph.
But as an indication of a trend they do seem kind of meaningful? This is indicating that revenue is growing rapidly. Divide by twelve if you'd rather see monthly revenue.
The only trend it indicates to me is that they continue to release completely unverifiable, disingenuous numbers intended to dazzle the incompetant who hear $30 billion and don't understand...
The only trend it indicates to me is that they continue to release completely unverifiable, disingenuous numbers intended to dazzle the incompetant who hear $30 billion and don't understand annuallized revenue.
It's mostly business customers. I think it has more to with Claude having the reputation of being the best for writing code.
I use it for my own work, but I think the reputation is currently overstated. The big three each have fairly similar performance when using their best models.
Just FYI, apparently Claude will sometimes reference Grokipedia...
Hey, I wrote that!
For the reference I didn't mean that to say that Claude is completely untrustworthy, but just like any AI you should actually check it's reasoning and ask it to find sources for its information. Claude is less sycophantic than other mega-sized LLMs that I've used but it's still there. When I tried asking it about local communities and we went back and forth on if a particular group was part of it - it said it was, I thought it wasn't and it agreed with me, then I did more research on it and it turns out that it actually was and Claude agreed with me again. You basically have to use it as a research assistant rather than a full-blown researcher.
As far as I can tell (and Claude backs me up on this), Anthropic doesn't have any business dealings with xAI or any of Musk's companies; it just brings up Grokipedia articles because it was the one that it came up with that seemed to fit the best. Heck, it might be getting picked because the articles are long-winded and have the appearance of authority.
Fair enough. I wonder if it’s possible to get Claude to exclude certain sources? I suppose I could check but… I won’t. I know me, I’ll forget by the time I close this tab.
Your profile has a “personal preferences” section which is global to all conversations. With Claude Code you can make an installation level or project level preferences file named CLAUDE.md. I think there’s something simelar for Claude Desktop, but that’s a step too far for me to try long enough to find out.
Edit: also forgot you can use Projects for the phone and web apps, which let you add in context in a similar manner, including using other types of documents like PDFs and images.
You can blacklist it in personal preferences on Claude. I just did that the other day.
I also immediately canceled and removed OpenAI after the pentagon nonsense. I switched to Anthropic and I am pushing my company to do the same.
I think Altman miscalculated by grabbing that opportunity. AI is kind of becoming a commodity. The difference between each companies flagship models isn’t huge.
As models plateau I suspect it’s going to become more about orchestration and efficiency.
IE having one model use other more efficient models as subagents for specific components of your request.
I did the exact same thing. I tested Gemini a little, first, but I'm disturbed by the amount of data I've given to Alphabet over the years, and I'd like to slow that bleed down.
Sticking to their values definately gave them good will with consumers. From a corporate standpoint, IT is more likely to build partnerships with the companies they use at home.
I mentioned this situation to my wife recently (she's too busy to watch all the tech drama in the world) and I'm pretty sure she swapped to Claude or will swap to it. She's using it in her business and when we talked about it I began to really realize that for her workload there is absolutely nothing tying her to one service; she deletes one app and installs another. Maybe Claude is better for coding and Grok is better at being a cringe edgelord, but these companies may have to start worrying about their reputation more.
Edit: She did switch to Claude and says its better in some ways but worse in others. Overall I think she's happy.
Honest question to anyone who knows these things: is Grok actually meaningfully better at any given metric than any other AI? I feel that it’s only ever mentioned because it’s going amok on X (which I am amazed is still culturally relevant, itself).
I also got rid of ChatGPT and settled on Claude exclusively after the Pentagon thing. But I doubt individuals move the needle on their revenue when I've heard of companies giving devs >$1k per month budgets for Claude.
Well, some of us, having chosen a favourite for personal work, choose that favourite among the options at their workplace (where they spend that >$1k per month budget).
You get a choice at work?! That's kinda cool, actually.
From the article:
[...]
This is quite the hockey stick chart: https://x.com/albrgr/status/2041288324464451617/photo/1
It's also a garbage chart of annualized revenue.
I'll be impressed when the show actual revenue next to expenses and not accounting magic tricks intended to hype gullible investors.
Hypothetical example:
I have an annualized revenue of $1 million. All I had to do was to project out my earnings this month which coincidentally included my parents dying and leaving me $10,000.
Another way to put this is:
In March, we earned an extra $2,500,000. If that's because a bunch of businesses bought new annual contracts, or price hikes, that's 1/12th the reported number.
This isn't an issue of annualized revenue, its an issue of accounting fraud where you're booking one-time sales as recurring revenue.
This situation is a combination of accounting fraud and Annual Contract Value here, not annualized revenue.
Source: I just read the Enron book, where the author goes through a variety of accounting frauds in great detail.
Okay but just because Enron did it doesn't mean Anthropic is doing it.
It definitely sounds like they are, or, at the very least, not being very honest in the way they're reporting.
E: Also, while perhaps not fraud, they said they calculate their annualized revenue by multiplying the last 4 weeks revenue times 13... and then report it right after the launch their fancy new models, a time when it's very likely that people are trying them out and then some will stay but many will cancel their subscriptions.
Yes, they're company-provided numbers that shouldn't be trusted as much as audited financial statements. But this isn't itself evidence that they're lying.
Yes, I don't necessarily mean outright lying a la Enron, but I do think everything they say should be treated with a heaping of salt.
I wasn't totally clear, but I agree with you. I don't have a reason to think Anthropic is committing accounting fraud. I was trying to say that the listed examples were of clear fraud, not legitimate caveats of annualized revenue numbers.
What is a magic trick but fraud with some razzle dazzle?
Annualized revenue does not mean they earned that much (since it hasn't been a year), so the big numbers don't quite mean what they seem. We are also missing the financial statements that would be available for a public company, so we have to go by these selective disclosures. Someone put them in a graph.
But as an indication of a trend they do seem kind of meaningful? This is indicating that revenue is growing rapidly. Divide by twelve if you'd rather see monthly revenue.
The only trend it indicates to me is that they continue to release completely unverifiable, disingenuous numbers intended to dazzle the incompetant who hear $30 billion and don't understand annuallized revenue.