Abhi Anand

@brave So your whole existence is because you can block YouTube ads?

PSA: You can block ads without paying money to the company showing you the ads in the first place.

@OfficialLoganK @CoelhoJvs @joshwoodward The voice system works much better, probably the best in ChatGPT, It's likely using a voice model to enhance output. I don’t want to rely on ChatGPT, but the voice feature forces me to. I prefer Gemini for its stronger reasoning and lower hallucination. Please fix this ASAP.

@CoelhoJvs Yes the team is on it, cc @joshwoodward

I’m hiring two [AI builders / Prompt engineers / Model whisperers] in SF, who should I talk to and consider?

@OfficialLoganK Tell me honestly, did you use Antigravity? If yes, which one did you use: Opus, Gemini, or both?

I spent the last week building what could easily be a $100M venture backed business… it’s truly wild how much leverage AI gives you.

@elonmusk Bro you are famous now, you got almost half million like

🤗

@Heccles94 Bubba's victim😂

Merry Christmas from ██████████████ and First Lady Melania Trump 🇺🇸

@Heccles94 Why his face hidden? Was he a victim of bubba...?

@bentossell You owe me 100M droid token before this year end. You might have forget but you had posted something earlier. I know it was of 10M only, but 100M will help me complete my project

he’s making a list and checking it twice gonna find out droids on it thrice

@bentossell @MohMaya97 Did you find any bug in your new product. Try droid

@MohMaya97 hospital is a boring place for hours on end. cuddles rotation ongoing don’t you worry

who's non-technical now?! also... confession the screen-recording was fully run by an agent that i was 'directing'. I didnt record myself doing the steps! i've built a mini system to script, use my computer, see what its doing, adapt, and film so i just have to record my money-maker and layer on top (+ time it 😬)

@OfficialLoganK And I am pretty sure all the benchmark is based on that.You are gud at general purpose which might give a good revenue but Google is known for innovation not to burn tpu to give better results.

@OfficialLoganK Bro, let me give you honest review, gemini 3 pro sucks in coding, it is lagging behind model like sonnet 4. The real power of ai comes to know when you do debugging, not just make some fancy single prompt project. 1/n

How long until AGI?

@bentossell Made with droid: github.com/abhioiitm/Web.…

no thanksgiving in the UK... so I'm giving out 1 Billion tokens! link a repo built with droid. First 100 people get 10m tokens each.

@bentossell Droid

@rork Rork

Claude Opus 4.5 is now live in Rork • The most advanced model from Anthropic: 80% on the SWE bench • Much better for mobile apps • 3x cheaper than Opus 4.1 Reply "Rork" to get 100 credits

@WeiboLLM Where can i test this model? If you havemdeplyed, give links

⭐ VibeThinker-1.5B — SOTA reasoning in a tiny model. 🚀 Performance: Highly competitive on AIME24/25 & HMMT25 — surpasses DeepSeek R1-0120 on math, and outperforms same-size models in competitive coding. ⚡ Efficiency: Only 1.5B params — 100-600× smaller than giants like Kimi K2 & DeepSeek R1. 💰 Cost: Full post-training for just $7.8K — 30-60× cheaper than DeepSeek R1 or MiniMax-M1. 🧠 Innovation: Powered by our Spectrum-to-Signal Principle (SSP) and MGPO framework. Model : huggingface.co/WeiboAI/VibeTh… Github: github.com/WeiboAI/VibeTh… Arxiv : arxiv.org/abs/2511.06221 #AI #LLM #Reasoning #OpenSource #SmallModel

@sundarpichai @bundeskanzler Bhai gemini 3?

@bundeskanzler Thanks for your support @bundeskanzler. Excited to make our biggest investment in Germany to date, as part of our ongoing commitment to Europe. We look forward to partnering with you to help German businesses and people compete and succeed in the AI era.

Wir bringen das Wachstum in Deutschland voran: 5,5 Milliarden Euro wird Google bis 2029 bei uns investieren – das sichert Arbeitsplätze. Unser Land ist und bleibt einer der attraktivsten Investitionsstandorte der Welt. Dafür arbeiten wir jeden Tag.

@OfficialLoganK @tylerwince @shresbm Now i know why gemini 3 is not getting released. Model needs much more improvememnt to to meet our expectation

@tylerwince good question, something we should look at @shresbm

We just shipped some nice Gemini API updates for developers using Structured Outputs. The API now supports: - $ ref for recursive schemas - anyOf union types - min + max numerical constraints - null types - property ordering adherence And much more!

@FactoryAI Please dm me any way to get 100M token on droid for free. It is amazing. Cant create multiple account

That one time at Droid camp…

@CyprianWetende @jeffwsurf @windsurf I had used it on vscode through copilot pro. I feel it overthink and then i have to clean up the mess with claude. For making a plan, it could be better.

@Supaulwale @jeffwsurf @windsurf I dont know man, but I just trust it with my complex work. Where are you using gpt 5 from? which tool?

We're a model company again. Coding models typically sacrifice speed over quality, which gives users a trade off when selecting which model to pick in @windsurf, but today, we release SWE-1.5, which outputs at a blistering 950 tokens per second, 13x faster than Sonnet 4.5. Not only that, it outperforms Sonnet Haiku 4.5 and GPT-5 on SWE-Bench Pro

@CyprianWetende @jeffwsurf @windsurf Forget*

@CyprianWetende @jeffwsurf @windsurf Gpt 5 is worse. It just start talking itself and forfet what to do. Worst model