Ritesh Mishra

[Ritesh Mishra @riteshmis]

What a privilege it was to witness Sam Neill through the eyes of a child

[Ritesh Mishra @riteshmis]

“One passenger requested and received medical assistance”

Imagine being sucked out of pressured cabin…. head first…at 16k ft and still having to request medical assistance

@grok how cooked is Ryanair legally?

https://t.co/nXH1vcmlol

[Ritesh Mishra @riteshmis]

Pretty dope! Curious to see if it can adjust grip while holding an object @BerntBornich?

[Bernt Bornich @BerntBornich]

Introducing NEO’s 25 Degrees of Freedom, tendon-driven hands — nearing or surpassing human-level dexterity, strength, speed, and reliability.

For seventy years, robotics worked around the hand problem. The humanoid bet is the reverse: it lives or dies at the fingertips.

[Ritesh Mishra @riteshmis]

For 250 years, freedom has had a place to call home.

God Bless America 🇺🇸

#America250

[Ritesh Mishra @riteshmis]

The internet became transformative when the cost of moving information approached zero.

AI becomes transformative when the cost of generating intelligence approaches zero (frontier open weight models)

I think @Apple has played the wise long game by not pouring billions into foundational model development

[Ritesh Mishra @riteshmis]

Open models + chip efficiency = the perfect storm

[ollama @ollama]

Gemma 4 is now nearly 90% faster on Apple Silicon with Ollama using MLX!

The speedup comes from improved multi-token prediction (MTP), now on by default for Gemma 4, with more models to come.

Ollama automatically tunes how many tokens to draft as it runs, so it never slows generation down when speculation no longer contributes to a speedup.

[Ritesh Mishra @riteshmis]

@elonmusk gonna release some benchmark metrics?

[Elon Musk @elonmusk]

To be clear, I’m not saying the Grok v9 foundation model will be mind-blowingly better than anything, but it will be a solid workhorse in the same league as Opus.

And the SpaceXAI cadence of model and harness improvement is speeding up tremendously, particularly due to a few dozen of the top Starlink/Starship engineers shifting a much of their time to AI.

Worth noting that the v8 foundation (Grok 4.3) model is a 0.5T model that finished training all the way back in December and has many fundamental flaws, so Grok 4.5 will feel like a gigantic upgrade!

[Ritesh Mishra @riteshmis]

Can’t wait to grow another appendage because someone vibe coded the containment protocols for a microreactor powering a “cloud” company with 20 employees and a WeWork office

[Office of Nuclear Energy | US Department of Energy @GovNuclear]

What excites you most about microreactors?

📸: @INL

[Ritesh Mishra @riteshmis]

Slippery slope, remember when Microsoft thought they owned the browser

[Shaun Ralston @shaunralston]

“Open source is very dangerous” @DarioAmodei testifies in this stunning video. The real danger isn’t open models; it’s @AnthropicAI convincing Washington that only they can be trusted with AI. Open innovation beats regulatory capture [1/2 👇]

[Ritesh Mishra @riteshmis]

@TechCrunch Where can I place a bet on the success of their prediction market?

[Ritesh Mishra @riteshmis]

@pmitu Limit reached resets 5pm (US/LA)

/upgrade to increase your usage limit.

[Ritesh Mishra @riteshmis]

@sflorimm Context windows got bigger than most conversations, and verification before output cleans it up.

If you’re doing serious code generation, though, you’re still talking about it. Code verbosity eats the window fast, hence steering files etc….

[Ritesh Mishra @riteshmis]

So @claudeai went down today 🗑️🔥

And agent traffic automatically routed to open models.

Service stayed online.

Marginal token costs dropped.

(Yes, yes… I already paid for the hardware 😅)