Elka Firmanda

Buzzer F-15EX

[SpaceX @SpaceX]

The Starship Recovery team has continued gathering imagery of the vehicle, which is still afloat in the Indian Ocean

[Flightradar24 @flightradar24]

23,075.92 km in 24 hours, 24 minutes.

We tallied up a few other facts and figures from this historic @Airbus A350-1000ULR flight.

[Elon Musk @elonmusk]

Interesting.

Grok 4.6 releases around August 7. This will be the 1.5T model with significantly improved SFT & RL.

Grok 4.7 will be the 2.1T model released a few weeks later. This will be better than 4.6 in every way, except slightly slower to serve, albeit with even better token efficiency.

[Liquid AI @liquidai]

Today we release LFM2.5-Encoder-230M and LFM2.5-Encoder-350M: bidirectional encoders that stay fast at long context, even on CPU.

> LFM2.5-Encoder-230M: about 3.7x faster than ModernBERT-base on CPU at 8,192 tokens. Under 30s per forward pass, versus over a minute and a half.
> LFM2.5-Encoder-350M: 4th of 14 models on GLUE, SuperGLUE, and multilingual classification, behind only three larger models, one of them nearly 10x its size.

🧵

[Elka Firmanda @ElkaFirmanda]

i need Q2_XXS quant please

[Md Ismail Šojal 🕷️ @0x0SojalSec]

The tiny Kimi-K3 that you can runs locally on a potato hardware.

- 2.8T to 0.18B.
- 0.10B activated.
- Same architecture.
- Same new attention design scale.
- Same DNA, smaller version
- compressed it into a 0.18B version for testing.
- now fits in 700MB.

You can actually load the new Kimi K3 architecture on normal hardware now.

[Qantas @Qantas]

The Project Sunrise @Airbus A350-1000ULR has touched down in Toulouse, France after completing a 24 hour 24 minute non-stop flight from Melbourne, Australia.

The aircraft flew northeast over the Pacific, over the US and Canada, then across the Atlantic and south over the UK to Toulouse, crossing three continents and two oceans, with two sunsets and two sunrises along the way.

[ModelScope @ModelScope2022]

Only 3B active parameters, yet SOTA agentic coding performance at its scale. 🚀 Meet KAT-Coder-V2.5-Dev, @KwaiAICoder's new 35B open-weight MoE coding model.
🤖 https://t.co/xKujahRrdz
📄 https://t.co/ICk2qaGwp4

🏆 Tops every reported same-scale benchmark: 69.4 on SWE-bench Verified, 63.0 on Multilingual, and 41.02 on Terminal-Bench 2.1.
⚡ Native 256K context, tool calling, thinking/non-thinking modes, and optional reasoning-history preservation.
🧠 Built on Qwen3.6-35B-A3B with 127K SFT samples + RL. Abnormal tool labels dropped from 9.34% to 0.28%, while continuous repetition fell to 0%.
📄 Open-weight and text-only—no vision components. Apache 2.0.

[Le Stagiaire de Hoonited @LHoonited]

De 1,65 à 2,5L/100km par passager.
La même conso qu’une 330D avec 4 personnes + bagages dedans.
Pour moi, c’est aussi impressionnant que le Concorde à son époque.

[Flightradar24 @flightradar24]

Touchdown! ✈️

After 24 hours and 23 minutes in the air, the A350-1000ULR has completed the flight from Melbourne to Toulouse.

[Colin McCarthy @US_Stormwatch]

This Airbus A350-1000 (Flight AIB35LR) is attempting the first-ever 24-hour flight by a commercial aircraft as part of a Project Sunrise test flight.

Now in its 17th hour, it is flying from Melbourne, Australia, to Toulouse, France, a journey of more than 13,500 miles.

[Madison Kanna @Madisonkanna]

So @waterloo_intern is unbelievably cracked.

you should follow him if you don't

[shaun @Biggest]

I do not agree with this. Thanks to the other employees who joined me in trying to push open-weight.

[Ahmad @TheAhmadOsman]

RTX 3090 owners trying to run Kimi_K3_3T_Q_0.0001_K GGUF

[Manan Gupta @yoitsmanan]

Introducing Neutrino-1: a new state of the art in intelligence per byte.

An 8-billion-parameter model that downloads in 2.56 GB,smaller than most 1B models, and runs on a MacBook, a desktop CPU, or a datacenter GPU from one artifact.

Open weights, Apache 2.0, today.

@fermion_ai

[Elka Firmanda @ElkaFirmanda]

wwaaaooooww

[Tibo @thsottiaux]

Back at the laptop. The usage limits have been reset for all paid users of Codex and ChatGPT Work. Weeeeeeeee. It's a good day!

[Alex Ziskind @digitalix]

unsloth already done with their Kimi K3 quant

[Teknium 🪽 @Teknium]

“we don’t support bans on open source models because that wouldn’t go far enough!”

[OpenRouter @OpenRouter]

GPT-5.6 Terra & Luna are 50% off exclusively on OpenRouter!

[the tiny corp @__tinygrad__]

Kimi K3 running on AMD MI350X with SGLang. Amazing work to @AMD @sgl_project @Kimi_Moonshot this worked great out of the box!