John Robinson

All is yours but for time and token...

[John Robinson @johnrobinsn]

@RaminNasibov alive and kickin

[PIYO @C_of_Creativity]

割り込みできる！すげー

会話型 AI 向けの 11B エンドツーエンドのリアルタイム音声フルデュプレックスモデルで、ストリーミング音声理解と音声生成を共同で実行します。

従来のカスケードスタック (ASR → LLM → TTS) とは異なり、このモデルは、1 つの統合アーキテクチャでフルデュプレックス、リアルタイム、シームレスな音声インタラクションを実現し、複数のモデルや API ハンドオフの必要性を排除し、エンドツーエンドのレイテンシを削減します。オープンで堅牢、かつ非常に自然な会話機能を提供することで、新たなベンチマークを確立します。

さらに、ツール実行中に自然な会話の流れを維持しながらツール呼び出しをサポートする初のオープンフルデュプレックスモデルです。各ツールに対して、LLM がツール呼び出しと応答をトリガーするテキストを生成するとすぐにエージェントによって話される特定の「保留」メッセージを定義できます！

https://t.co/opNYDMbRQ2

[John Robinson @johnrobinsn]

Lose Yourself

[Michael Warburton @For_Film_Fans]

In 1985, the famous classical pianist Horatio Gutierrez dropped out of appearing on THE TONIGHT SHOW at the last minute, JOHNNY CARSON invited a member of the audience to take Gutierrez’s place.

Up stepped a guy called DAVID TOLLEY…

[Hunter Weiss @Hunter_Weiss]

Avidrone is now in 1st place

29.3 lb drone, 112.4 lb payload (3.84:1)

Course completed in 19 minutes 17 seconds

[John Robinson @johnrobinsn]

@M44_1RJ @elonmusk nah we all know it's going to be $420.

[mrdoob @mrdoob]

Transferring data with rapidly flashing qr codes 🫰
https://t.co/3qnijxjV1R

[John Robinson @johnrobinsn]

Kinda like shark scales 🔥

[eXnihilo @eXnihilo448]

@DJSnM 🤷‍♂️I know it sounds crazy... If the leading edges of the tiles were hidden a little bit by the trailing edges, it would change everything. If the tiles were slightly tilted forward, like 3°, it might be enough. Not scales, but just ripples...@elonmusk

[Arthur MacWaters @ArthurMacwaters]

Humans for scale

[John Robinson @johnrobinsn]

Open-source distillation of closed models isn’t theft. It’s the closed models finally paying rent on the public data they trained on. All of it was distilled from humanity to begin with.

[Jürgen Schmidhuber @SchmidhuberAI]

I support open-source models distilling what commercial companies distilled for free from the entire internet. I published distillation for free in 1991 in Europe - this was copied in the US and in China (https://t.co/mddh8XmfAs)

[Jürgen Schmidhuber @SchmidhuberAI]

I support open-source models distilling what commercial companies distilled for free from the entire internet. I published distillation for free in 1991 in Europe - this was copied in the US and in China (https://t.co/mddh8XmfAs)

[Anderson Mancini @Andersonmancini]

Live link: https://t.co/ieY5UD5rB0

Built to showcase what's possible with real-time WebGPU rendering using #threejs. Just a 6.3 MB download, yet the lighting looks baked—until you grab the sun and watch everything update in real time.

I hope you like it 🤗

[John Robinson @johnrobinsn]

Elon: We are in the Singularity

Singularity: Hold my beer...

[Elon Musk @elonmusk]

We are in the Singularity

[Starlink @Starlink]

High-speed internet from space for the future of autonomous vehicles 🛰️🚗🤝

[𝙺𝚒𝚖𝚋𝚊𝚕 𝙼𝚞𝚜𝚔 🤠 @kimbal]

I’m so proud of our Nova team for flying incredible shows for FIFA this year.

This is my favorite shot from our closing show in NYC this weekend.

[Linus ✦ Ekenstam @LinusEkenstam]

This is how football should be visualized.

Spain vs. Argentina paints a clear picture

The graph shows real momentum, rendered in 3D, supported by data on goals, shots, xG and more.

Spain are world champions again,
Ferran Torres made sure of it.

Graph by Intelligoal

[Dudes Posting Their W’s @DudespostingWs]

I don’t think people realize how absurd this is. Scottish runner Josh Kerr just broke the mile world record in 3:42.66, an average speed of 16.2 mph. Most people couldn’t even briefly reach that speed, let alone sustain it for an entire mile.

[John Robinson @johnrobinsn]

Slowly at first... then all at once...

[Gaverick @eroshelli1]

@SawyerMerritt @Tesla Saw some in Tyson Corner Virginia today👀 Probably nothing.

[Dave @dae5id]

thanks china

[KyzoroX @kyzoroXX]

Chinese factories are buying RTX 5090s in bulk, desoldering the chips, and rebuilding them as 128GB server cards. Roughly $4,000 each.

Same money as a DGX Spark. Same 128GB. Six times the bandwidth.

The reason this exists: export controls block Chinese labs from buying dedicated AI chips, so consumer gaming cards became the workaround. Rip the GPU and memory off the board, solder onto a custom server PCB, add a blower cooler so it fits a rack instead of a gaming case. Automated soldering lines, robotic arms. Started with the 5090, now spreading to the 5080, 5070 Ti, 5060 Ti.

NVIDIA gets paid either way. They just lost control of what happens after the sale.

Two things follow from this:

Gamers are the collateral. Consumer cards are being bought in volume and stripped for AI, which is part of why you can't get one at a sane price.

And the "$4,000 for 128GB" pitch that DGX and Strix Halo are built on now has a grey-market competitor with dramatically better bandwidth — assuming you can get one, trust the soldering, and live without support.

The most interesting AI hardware of 2026 isn't being announced at keynotes. It's being reflowed in a workshop.

[shadcn @shadcn]

few weeks ago, Fable 5 was so advanced it needed the gov and had to be taken offline.

today, we have access to arguably better models for $20/month.

took what? six weeks?