Chris Tar

(* Compacting conversation...)
🍌 Nano Banana & Image modeling @googledeepmind

[Arena.ai @arena]

Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) has entered the Text-to-Image Arena. It ranks #5 overall.

Google's latest Gemini-3.1-Flash-Lite-Image lands on the Pareto frontier: scoring 1251 at just $0.034/image. This is near flagship quality at a fraction of the price.

The cost-to-quality curve just got a lot more interesting.

Congrats to @GoogleDeepMind on the release!

[6529 @punk6529]

total and complete victory for Kurzweil

called it all 35 years ago and never wavered

[Chris Tar @ctar]

@espn What is the deal with the audio artifacts on the local announcer...not just tonight. Are the announcers working from their laptops at home over Skype?

[Jeff Dean @JeffDean]

This is absolutely shameful. Agents of a federal agency unnecessarily escalating, and then executing a defenseless citizen whose offense appears to be using his cell phone camera. Every person regardless of political affiliation should be denouncing this.

[Chris Tar @ctar]

Amazing infographics, posters, slides generation based on your content, in NotebookLM today (via Nano Banana Pro)

[NotebookLM @NotebookLM]

Infographics & Slide Decks are 100% rolled out to all Pro users!

One of our favorite examples that came out of early testing was uploading your resume/LinkedIn and creating a custom, visual representation of your career. Here is our social media manager's:

Share yours too!

[Sundar Pichai @sundarpichai]

You went 🍌🍌 for Nano Banana. Now, meet Nano Banana Pro. 

It’s SOTA for image generation + editing with more advanced world knowledge, text rendering, precision + controls. Built on Gemini 3, it’s really good at complex infographics - much like how engineers see the world:)

[Kaushik Shivakumar @19kaushiks]

Nano Banana Pro marks a significant jump in accuracy of rendered text within images across many languages.

[Mostafa Dehghani @m__dehghani]

All calm. For now.

[Phillip Lippe @ICML 2026 @phillip_lippe]

Gemini 3 Pro is out with large jumps in multimodal understanding and reasoning. Sounds useful for another application we're picturing... 🎨

[Chris Tar @ctar]

Nano Banana Pro is out with significantly improved text, diagram and infographic capabilities - Search-driven generations mean how-to's for many topics

[Google DeepMind @GoogleDeepMind]

Our latest image model has enhanced reasoning from Gemini 3. 🧠

It can connect to @Google Search's vast knowledge base to help visualize anything using real-time information - from an infographic for a recipe, to an illustrative physics diagram.

[Radu Soricut @RSoricut]

When you’re done studying and want to catch up on your favorite team, you can just ask “Infographic of last Rams game”. Having Google Search ground your queries and Nano Banana Pro show them visually is seeing two great technological achievements acting in perfect tandem:

[Nicole Brichtova @nbrichtova]

Heard some people have been waiting for this one. Meet Nano Banana Pro!
We kept the 🍌 and upgraded everything else.

https://t.co/zWReO0QyKi

[Nicole Brichtova @nbrichtova]

We just released the world’s best image model. Not bad for a fruit. 🍌

[Demis Hassabis @demishassabis]

The new Gemini 2.5 image model🍌is by far the best out there with a whopping +180 ELO point lead in image editing & it really excels at character consistency.

Available for free in the @GeminiApp right now. Try uploading an image & playing around with it, it's pretty amazing!

[Chris Tar @ctar]

@_ZachGriff @united @Starlink I flew a Hawaiian flight that had Starlink, I wonder if they solved this or if it's a hidden issue for them as well?

[Tu Vu @tuvllms]

🚨 New paper 🚨

Excited to share my first paper w/ my PhD students!!

We find that advanced LLM capabilities conferred by instruction or alignment tuning (e.g., SFT, RLHF, DPO, GRPO) can be encoded into model diff vectors (à la task vectors) and transferred across model versions.

💡You don’t necessarily need to fine-tune from scratch again for every new base model version. Instead, fine-tune once and add the diff vector to updated versions! ♻️♻️♻️. This can also offer a stronger and more computationally efficient starting point when further training is feasible.

📰: https://t.co/q2pPZcHGdY
More 👇

[Chris Tar @ctar]

@lateinteraction The way I’ve been thinking about this and explaining (LLM-based) agents is simply an acknowledgement or realization that assisted/guided task decomposition allows you to solve problems that otherwise aren’t easily solvable with LLMs

[Arena.ai @arena]

Massive News from Chatbot Arena🔥

@GoogleDeepMind's latest Gemini (Exp 1114), tested with 6K+ community votes over the past week, now ranks joint #1 overall with an impressive 40+ score leap — matching 4o-latest in and surpassing o1-preview! It also claims #1 on Vision leaderboard.

Gemini-Exp-1114 excels across technical and creative domains:

- Overall #3 -> #1
- Math: #3 -> #1
- Hard Prompts: #4 -> #1
- Creative Writing #2 -> #1
- Vision: #2 -> #1
- Coding: #5 -> #3
- Overall (StyleCtrl): #4 -> #4

Huge congrats to @GoogleDeepMind on this remarkable milestone!

Come try the new Gemini and share your feedback!

[Logan Kilpatrick @OfficialLoganK]

Gemini is now accessible via the OpenAI libraries! Update 3 lines of code and get started with the latest Gemini models : )

https://t.co/fG2f6EFsFY

[Chris Tar @ctar]

https://t.co/Qp77l6qxGX