I build AI agents in public · Cognition Ambassador · Demos + lessons
Joined February 2025
- Tweets12.7K
- Following2.4K
- Followers1.2K
- Likes29.8K
Florin retweeted
Elon just said Grok Bot cost too much to serve.
Meanwhile Devin runs Linux, Mac, and Windows.
Look at the specs. 32GB RAM vs Grok Bot's 16GB.
SWE-2 is way better than Grok once you count price. 11x cheaper per task. And you pick the model.
Devin's subscription is underrated.
Florin retweeted
Devin spent 36 hours and ~$2k getting a model running cheaply on Trainium, work that used to take specialist teams months.
Serving AI on more available hardware is becoming more practical now that agents can cheaply automate porting and optimization work that use to require specialist teams. ⚡
Florin retweeted
How do you hide the fact that you’re halving the credits? By renaming Max 20x to Ultra. It’s 10x now. Wow.
They got me at first, but this is super misleading. If Anthropic is trying to build trust, this is definitely not the way.
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: claude.com/pricing
Florin retweeted
The @corridor team has deep expertise in AI code security, congratulations @AshwinRamaswami @jackhcable and welcome to Cognition.
Security has always been a top priority @cognition, this strengthens our position as a leader in the space.
Today, Cognition is acquiring @corridor.
Securing the world’s software is the most important problem in technology today.
Let’s get to work!
Florin retweeted
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Florin retweeted
Had a nice dinner with @imjaredz and the @DevinAI @cognition crew last night.
Couldn't live without Devin at @HelloUntangle.
Switch to a great cloud harness.
If you're doing local dev, you're working in the past.
Try it and you'll see the light.
Florin retweeted
FrontierCode 2.0 includes broader coverage of tasks, drawn from more repositories and spanning more categories, including security patches and performance optimizations.
Every Devin plan gives you access to most models on the FrontierCode 2.0 leaderboard, including Claude, GPT, Gemini, Grok, SWE, GLM, DeepSeek, and Kimi families of models.
Introducing FrontierCode 2.0: our new coding benchmark that measures a model’s ability to go beyond the spec.
Most coding evals measure a model’s ability to write code that meets a spec. With FrontierCode 1.0, we asked a new question: would you actually merge the PR?
FrontierCode 2.0 is comprised of more diverse tasks, featuring larger scopes and higher mergeability standards. We are now asking: not only would you merge this PR, but should you?
One key feature of FrontierCode 2.0 is underspecified prompts. Rather than including every engineering decision, prompts are deliberately ambiguous, giving the models room to demonstrate their judgment.
The result is a benchmark that is far from saturated, with the top model scoring just 45%.
Florin retweeted
Introducing FrontierCode 2.0: our new coding benchmark that measures a model’s ability to go beyond the spec.
Most coding evals measure a model’s ability to write code that meets a spec. With FrontierCode 1.0, we asked a new question: would you actually merge the PR?
FrontierCode 2.0 is comprised of more diverse tasks, featuring larger scopes and higher mergeability standards. We are now asking: not only would you merge this PR, but should you?
One key feature of FrontierCode 2.0 is underspecified prompts. Rather than including every engineering decision, prompts are deliberately ambiguous, giving the models room to demonstrate their judgment.
The result is a benchmark that is far from saturated, with the top model scoring just 45%.
Florin retweeted
The world's most powerful cyber models, including Claude Mythos, are available in Devin Security Platform. We're also offering hundreds of millions of dollars in credits to use it.
If you want to defend faster than attackers, you have to find and fix flaws at machine speed. That’s why we’re building the best AI-powered security platform in the world.
Devin already helps engineering teams write more secure code, now it can secure the code they already have.
Florin retweeted
We rewrote Devin Desktop to use the same design and rendering system as Devin Cloud, while preserving full local agent compatibility.
It took millions of tokens, but the result is worth it: the UX is 10x better than Windsurf used to be.
Florin retweeted
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
Florin retweeted
I tried the Devin Next desktop app, and the gap between it and the main app is huge.
The main app used to crash constantly once I ran a lot of agents, so I kept falling back to the CLI. It also opened windows everywhere, which made staying organized almost impossible. The whole experience felt scattered and in need of serious polish.
Devin Next fixes most of that. I ran 40 SWE-2 Max agents at once, each researching and writing a complex 5,000-word article on a different topic. Zero crashes. Completely smooth.
Everything is easier to reach now. Features that used to feel buried are right where you expect them, and the whole app feels cohesive and intuitive.
Goal Mode is finally here. On very long tasks, Devin would sometimes stop to ask if it should keep going. Goal Mode is a game changer.
What I'd still like to see:
A weekly usage indicator next to the context button
Arabic language support
Full MCP support for tools like Gmail and Figma, which don't fully work yet
Devin Next is a great experience. Download it and start using it, you'll enjoy Devin way more. November is going to be an exciting month for Devin.
Florin retweeted
We’ve open sourced an agent plugin that helps secure systems against recent cybersecurity threats.
You can install them with any agent or use them in Devin today:
devin.ai/blog/vulnerability-…
Florin retweeted
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
Florin retweeted
Devin Fusion. Better and cheaper.
Devin Fusion has claimed a top score of 68 on the @ArtificialAnlys Coding Agent Index with an Opus 5.5 lead and GPT-6 Luna sidekick.
Devin Fusion outperforms Claude Code with Opus 5.5 on both performance and price – earning a score of 68 at a 31% lower cost per task.
