One

@bualtbicha @Nathenal_ First time seeing my profile picture

@Nathenal_ ክርስቲያን ሰው አይጠላም፣ የደከመ ሰውም አይመለከትም 😂😂

You're a sick person if you don't hate comedian Eshetu enough

@AD_CFCET @BadheeA There is always "justification" to do such atrocities. The only good way is not to do it. It's their culture that is barbaric and causes this. If we do the same in Ethiopia with some kind of "our justification" we are no different and cant complain. But yes, we need some action

@BadheeA Sad to watch. Can we burn down that Mandela house at Gerji Mebrat which inspired by Nelson Mandela. I don't think the owner is an Ethiopian

Two Ethiopian 🇪🇹 shop owners in South Africa 🇿🇦, robbed and gunned down by anti immigrant protesters, Caught on CCTV camera. The situation seems beyond robbery and protest. The two guys were surrendered and lying on the ground when the young man first check for the environment and mercilessly shot them down. There must be a terrorist insurgence and therefore the government must act to prevent further harm.

@mehran__jalali DM me

(if you're seeing this and you work at anthropic and want to give me credits to run some benchmarks, my DMs are wide open!)

@Google @arena Useless. The only edge for the flash lite models was the pricing, google seems to have decided to lose that whole userbase.

Gemini 3 Flash-Lite by the numbers: ✅ 1432 Elo score on @arena Leaderboard ✅ 86.9% on GPQA Diamond (scientific knowledge) ✅ 76.8% on MMMU Pro (multimodal reasoning)

Developers can now preview Gemini 3.1 Flash-Lite, our fastest and most cost-efficient Gemini 3 series model yet. With a 45% increase in output speed, it outperforms 2.5 Flash and features dynamic thinking levels to match task complexity. Rolling out in preview today in @GoogleAIStudio and Vertex AI.

@OfficialLoganK @GoogleDeepMind Completely useless. The only edge for the userbase of flash lite models was the pricing.

Introducing Gemini 3.1 Flash-Lite 🔦, a huge step forward on the boundary of intelligence, beating 2.5 Flash on many tasks.

@OfficialLoganK @thegenioo @GoogleDeepMind Its useless bro, the only, single edge to use flash lite models was the pricing.

@thegenioo @GoogleDeepMind very strong model!

@AiBattle_ Color choice seems from OpenAI

A new model “Galapagos” is currently being tested on DesignArena The model has a front-end style similar to GPT-5 models Might be a new GPT model

@ai_for_success Just lost respect for this account

Nik Pash was fired by Cline for making racist comments about Indian coders and is now hired by OpenAI, working on the Codex team. OpenAI, you have lost all respect.

@sama @thorstenball @sama SPEED SPEED SPEED Truly, FAAAR better and thorough than Opus 4.5 for development (5.2 high), if it wasn't for speed. incomparable

@thorstenball what's been working well, and what would you like to see us improve in 5.3?

man, GPT-5.2 is very good.

@dobsolute had to follow after this post

I think 99% of people who are less than 2 years into tech should just quit and pivot to something else (ofc this refers solely to wagies). Generally, the ability to negotiate a good salary depends on the scarcity of people with your skill set. In a market where supply greatly exceeds demand, unless you are truly exceptional there will be many of you, giving your boss all the leverage when hiring. From there he can make all kinds of demands (less pay, worse conditions, etc.) In other words what incentive would your boss have to meet your conditions when there's a flood of unemployed talent dying to work for peanuts? The rest 1% of people (the exceptional ones) will probably be fine for a while, but it's uncertain if even they will hold any leverage in 3-5 years. On the bright side, instead of playing the game with such shit odds, you can pump out your own shit, it's easier than ever.

@demishassabis @demishassabis How about Gemini 3 Flash-Lite? 🥹

We had a pretty productive year! 🚀

@OfficialLoganK @WBlueu You should try GPT 5.1 high. Never misses a thing you say and feels responsible

@WBlueu I used to ask for a few features in a prompt, now I just ramble on and on, then ask the model to add 5 more things

The way I use AI has changed with Gemini 3… I used to try and perfect the minimal ask to get the model to be successful at a task. Now, I’m constantly pushing myself to be more ambitious, asking for 5x more than I did before with a single prompt. And Gemini 3 crushes it.

@thsottiaux * Better verbosity and communication * extensive shortcut keys (without replacing core vscode ones) * better keyboard based ux

Codex team is working on a few experimental projects that are starting to shape up and I’m excited to share more about soon. But I’m curious, what would you like to see ship or improved by the end of the year other than better models?

@osanseviero All we need. What makes claude models best is instruction following, every single phrase. x.com/powfowab/statu…

We launched Gemini 3 a week ago and the community response has been incredible! 🚀 We're listening: What features or capabilities would you like to see next? Drop your wishlists below👇

@alisa_fortin This is the first bad image I saw from nano banana Pro. It doesn't seem to have spatial understanding. The character Yoda is facing us, and at the same time looking and lifting an item behind it. The irony is that I saw the first ever bad look from a Google team member.

@zaveng @OfficialLoganK That's why the general model should not lose its edge being tuned for coding, nor bad at coding being good at general tasks like now. A separate coding model like codex makes everyone happy

@powfowab @OfficialLoganK not everyone needs coding.

Gemini

@OfficialLoganK @simpsoka voted "Keep as is" 😁

@OfficialLoganK Gemini 3 pro, although capable, is too general for not as very SOTA for coding. I believe if Google released a coding Gemini model, it would be God tier. Let's ask google via poll to show what we want, mention deepmind engineers too

@OfficialLoganK @MatthewBerman Gemini 3 code when? Obliterate all models

@MatthewBerman pretty tired too tbh, need to be back at office at 5 AM, the cost of relentless progress :)

@OfficialLoganK @OfficialLoganK what are your thoughts on this? Could be good, bad...