RDP

Solutions Architect and an off day Philosopher, Photographer, History Buff 🏔 Likes/RTs ≠ Endorsement

[Shahir @TslShahir]

We have probably the fastest TTS ever on device. Built on Embedding Gemma from @googlegemma . Tested on @sabeshbharathi 's M5 pro. 300x realtime

[RDP @Utpalodhi]

@ShivAroor Times of India 🤡

[RDP @Utpalodhi]

@sunilchengalath @HelleLyngSvends Tu ek chomu kutta hai, tujhe soch ne ki zaroorat nahi

[RDP @Utpalodhi]

@salilnair1 @RoshanKrRaii Kahan? mujhe toh bas 4-5 chomu kutta dikh rahe, ye patriots kahan milenge?

[Andrew M. Dai @AndrewDai]

Spent years watching models ace a single image and completely whiff on anything that moves or changes angle. Today we’re fixing that.

Try Glimpse now, our first visual thinking model with competitive frontier performance compared to Astra on key public visual reasoning benchmarks.

[0xSero @0xSero]

People with Nvidia RTX 30 series chips rejoice!

You have Qwen3.8-Next-Flash

- 25 to 60 tok/s decode
- 610 to 2100 tok/s prefill
- exl3-2bpw for 3060/3070
- exl3-3bpw for 3080/3090

This is the cheapest, smartest, fastest thing you'll be able to run.

[GameUnionTV @GameUnionTV]

LMAO, #Blender is finally free and open source!

[Nandakishor m @Nandakishorm1]

vega vs jev vs laya. Btw, Vega can run on a CPU, your smartphone, browser on the edge. With a speed of 75 ms in apple M5 16 GB

[Tencent AI @TencentAI_News]

A 3B active model beats Opus 5 at memory. We open-sourced the full stack — model, engine, and app.

Hy Memory handles chat, work, and code memory in one layer, and it's first on all four benchmarks we ran, leading MemoryArena by 10.2 points while cutting memory volume 51.8% and recall tokens 39.3%.

The model isn't tied to our engine. Drop Hy-Memory-30B-A3B into Mem0, MemOS, or EverMemOS and it still comes out on top, even on the one system it never trained on. It's already running in our own products, where one assistant's multi-hop recall went from 50% to 90%. It also unifies memory across Codex, Cursor, WorkBuddy, and more.

Repo: https://t.co/M7PkOJVyE5
Weights: https://t.co/U04mNgmVuS

Questions welcome in the replies.

[Sann @san_x_m]

Yesterday in office, a few of us were talking about Navratri and Durga Puja. Somehow it went to food.

Couple of guys from northern part were proudly sharing how strictly they fast. No onion, no garlic, full nine days.

Then everyone looked at the Bengali guy in the group, from Tripura.

I felt a little awkward. I know Bengali culture a bit and didn't want him put on the spot.

He smiled and said calmly,

"For us Bengalis, Maa Durga is our daughter. She comes home once a year. When your daughter comes home, you don't fast. You feast."

[RDP @Utpalodhi]

@Indianomy_ @RahulGandhi Areh kutta

[liyong @liyong2077]

60fps ZeldaTearsOfTheKingdom Recompiled To Mac， thanks @happyjoejiang ！！！

[🚨Indian Gems @IndianGems_]

To save the image of the Rupee from falling to 100, the RBI has spent half of its $140 billion FX capital haul: Reuters

[Qwen AI Fans @QwenAIFans]

you don't need a 128GB MacBook for Qwen3.8-Flash-Next anymore

Strata runs the 125B model on a 12GB gaming GPU + 32GB RAM.
RTX 5070: 94 tok/s at Q2, 53 tok/s at IQ3_S

Q2 is a heavy quant, so check quality on your own tasks before you trust it. still wild. 25K stars on GitHub already

[Nic Cruz Patane @niccruzpatane]

HERE IT IS. The Next Generation Roadster developed by SpaceX and Tesla!

There is a fan-suction system at the bottom of the car that can pull up to 2 gs allowing it to stick upside down.

It just flipped UPSIDE DOWN with cold gas thrusters and landed. HOLY SHIT.

[RDP @Utpalodhi]

@Erorr8653 @smitadeshmukh @BBCWorld Tujhse kisine nahi pucha dehati kutta

[Tesla @Tesla]

Real life, baby

[AI at Meta @AIatMeta]

Today we’re releasing Muse Spark Open, the most capable US open-weight model to date.

Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.

Key Model Specs:
→ 816B total / 42B active parameters
→ 1M token context window
→ Natively multimodal across text, image, video, audio

[Daanish Khazi @bertgodel]

We are releasing Ultramarine 1 Flash, a frontier-class model built for knowledge work. It is open-source and achieves similar performance to Fable 5, while being 60x cheaper.

It was post-trained by a team of 5 for ~$300k all-in. Available on API now, with weights releasing in 2 weeks.

[Hermann Kumbong @KumbongHermann]

🚀 Introducing Next-Concept Modeling (NCM): our new paradigm for fast, parallel, coarse-to-fine text generation.

Compared to strong diffusion language models (DLMs) at similar parameter size, NCM achieves:

⚡ Up to 31× faster generation on downstream tasks like summarization and machine translation with comparable or better quality.

🏎️ 10–21× faster unconditional generation across TinyStories, LM1B, and OpenWebText, with comparable or better generative perplexity and using 3–5× fewer training tokens.

🧩 Better performance on reasoning tasks like Sudoku, outperforming both AR and diffusion baselines with fewer sampling steps.

With @ishanskhare, @IddahMlauzi, @ekellbuch, @simran_s_arora, @Azaliamirh, and @HazyResearch.

More details in the thread 🧵👇