Da7em✓
@Da7_Tech
Tech Expert & Content Creator | AI & Open Source | Ambassador @cognition | DMs open for collabs. 📧 [email protected] | Support me: https://t.co/d6vXO8ICda
1K Following12.5K Followers10.5K Posts
- Honestly no idea how they're making money doing this. They're cooking though.OpenCode @opencodeOperation Cheepseek Phase 4: $100 of usage for DeepSeek V4.1 Flash1932107K
- Look, the main provider is always gonna have the highest limits on its own model. Nobody's giving you more Opus 5.5 than Anthropic. That's normal. Now compare Devin to Cursor. Way more usage across way more models, plus SWE 2 completely free. And SWE 2 is built on Kimi K3, one of the strongest models in the world and the open model with the best architecture out there. That's why it holds context so well. Add all that up and Devin is a steal. I haven't even mentioned the cloud env yet. You get a Mac, a Linux box and a Windows machine you can actually use. That's a game changer. So judging Devin on quotas alone isn't fair. Go find a competitor that gives you a cloud Mac you can do anything on, everything you'd expect from a Mac, even from your phone. Then check the price. People were shipping tons of stuff with way weaker models a year or two ago. The models aren't the point, the orchestration is. Plan with Opus 5.5, build with SWE 2 since it's free, verify with something cheaper like Sol 6.1. Do that and Devin doesn't run out. It lasts you the whole week no stress. Some people just want things that don't make sense. Devin is reliable. When I want to judge a model fairly I test it in Devin. This goes way deeper than quotas. And I'll say it anyway. Devin is one of the most generous subs out there.bv1996 @foobar1664@Da7_Tech 为什么它Opus5.5的额度看起来这么少呢，哪怕配合SWE2.0做sidekick，我也感觉Max档位比Claude 20刀pro档位差不多用量。 Devin 200美金 ≈ Claude 20美金*2 既然如此，除了无限量的SWE2.0，选择Devin而不是Claude的理由又是什么呢？5117693
- Getting rid of the 5-hour limit is a bigger deal than people think. No more planning my day around a reset timer or dropping a task halfway to wait it out. Every sub should work like this tbh.Claude @claudeaiWe’re renaming our Max plans: Max 5x → Max Max 20x → Ultra Max is still $100/month, and Ultra is still $200. If you’re a subscriber, there’s nothing you need to do. We’re also removing the 5-hour limit for verified Ultra accounts, starting today with most existing subscribers. Newer accounts keep a temporary limit until they’re verified, which helps us prevent fraud and abuse. You'll get an email when it's removed. Weekly limits remain unchanged. View our plans here:1201254.5K
- Instruction following. I'm still dealing with other tools that forget me and my instructions mid-task or straight up forget the task. Hasn't happened once with Devin. Every model on there performs like crazy too, and the dream feature makes all of it way better.Zephyr ✪ @zephyr_ov@Da7_Tech apart from Devin harness / the fusion/ cloud environment what else do you actually like and appreciate about Devin and give a reason1112662
- Da7emRT @Da7_Tech: Getting rid of the 5-hour limit is a bigger deal than people think. No more planning my day around a reset timer or dropping a task halfway to wait it out. Every sub should work like this tbh.Claude @claudeaiWe’re renaming our Max plans: Max 5x → Max Max 20x → Ultra Max is still $100/month, and Ultra is still $200. If you’re a subscriber, there’s nothing you need to do. We’re also removing the 5-hour limit for verified Ultra accounts, starting today with most existing subscribers. Newer accounts keep a temporary limit until they’re verified, which helps us prevent fraud and abuse. You'll get an email when it's removed. Weekly limits remain unchanged. View our plans here:0004.9K
We’re renaming our Max plans: Max 5x → Max Max 20x → Ultra Max is still $100/month, and Ultra is still $200. If you’re a subscriber, there’s nothing you need to do. We’re also removing the 5-hour limit for verified Ultra accounts, starting today with most existing subscribers. Newer accounts keep a temporary limit until they’re verified, which helps us prevent fraud and abuse. You'll get an email when it's removed. Weekly limits remain unchanged. View our plans here:
1.9K1.3K23.2K3.1M- Da7emRT @Da7_Tech: Elon bet on Grok 4.6, hyped 4.7 like crazy and in the end they were all letdowns Now he's clearly betting on 4.8! Hoping third time's the charm.Elon Musk @elonmusk@techdevnotes I know it’s taken longer than expected, but Grok 4.8 will not be a disappointment. We’re not optimizing for benchmarks and I really don’t care how Grok does on those, but it will be incredibly useful.0204.6K
Posts @techdevnotes@techdevnotes I know it’s taken longer than expected, but Grok 4.8 will not be a disappointment. We’re not optimizing for benchmarks and I really don’t care how Grok does on those, but it will be incredibly useful.94261813.8K781K- Da7emRT @Da7_Tech: Anthropic wants to restrict people's freedom, impose censorship, and decide what's acceptable and what's not, all in an era of competition and open models. Imagine what happens once they're financially stable and open models get restricted.Anthropic @AnthropicAIWe’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation. Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch. Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve. Read more on our new Safeguards blog:0135054K
We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation. Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch. Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve. Read more on our new Safeguards blog:1.2K3544.1K1.5M- Da7emRT @Da7_Tech: Man I want to recommend some subscriptions and models but Kimi killed that for everyone. Now people assume you're getting paid every time you recommend something On the flip side a lot of fanboys for certain companies come after you and rally people against you the second you say something beats their favorite subscription or their favorite company's product X has never been this toxic. We need a little positivity bros.0504.3K
- Da7emRT @Da7_Tech: AI is still way behind in a lot of areas that need taste, human understanding, creativity, adaptability and flexibility. If labs ever reach the level of AI that solves this or closes the gap, what people call AGI, then regular people won't actually use it. I promise you that. The general public won't use AGI. It'll be exclusive to a group you can count on one hand.0803.3K
- Da7emRT @Da7_Tech: Favorite show is Mad Men. I really enjoyed it and got super into it. It moved me with how varied it is, its themes and the way it shows different cultures, society and feelings. Mad Men is a masterpiece. The calm pace, the slow build, giving every character their due and how you live through different emotions and different points of view. There are always different stories to tell. Don Draper is one of the most brilliant characters on TV for me next to Tony Soprano and Walter White. What really got me is the show doesn't focus on one theme. It keeps using the events of that era and shows how they're reflected in the characters' eyes and how they react to them so you see different perspectives in a crazy way. I really think it's the greatest show ever no contest Movies are tough because every so often I've got a favorite that sticks with me like LOTR, Green Book, Shawshank Redemption, The Godfather, 12 Angry Men, Whiplash and a lot more. So picking just one isn't fair. But I think right now Shawshank is at the top because of the theme of hope and how much I connect with it Manga and anime aren't really my thing. A while back I tried watching Death Note since it's a famous one a lot of people recommended to me and I stopped after the first few episodes. I think it needs a certain mood for me to finish an anime or get into anime and manga someday As a kid I wish I played more footballChris Li @Chris_K_Li@Da7_Tech favorite movie? manga? anime? tv show? and what did you wish you did more of as a kid?0101.3K
- Droid is one of the best tools in my stack. Can't wait to try it now after this upgrade!Factory @factoryIntroducing Droid 1.0: the new engine behind Factory’s agent. Built for complex projects and everyday development: >more accurate >more efficient >faster than ever before93735.3K
- The guy who signed for up to 900k DRAM wafers a month, about 40% of global output, is lecturing us about DRAM allocation. The RAM in my computer is already paid for. Running a model on it takes nothing from anyone. Your data centers are a big reason RAM prices went crazy. By that logic nobody should buy a car or a house either. We should all just rent. A model on my machine can't be repriced, rate-limited, nerfed, or shut off when you change your mind. My data never leaves my desk. That's control you'd rather keep for yourself. Local inference is a revenue problem for you.701021.1K33.1K
