erik

21. local llm & agents with agency. currently @ defence/robotics startup 🇨🇭

(follow for AI takes and some shit posts)

[erik @0xerik]

@simcomsystems Hmm, does mine annoy you?

[erik @0xerik]

@Laz4rz Did you end up getting it? Thinking about it rn

[erik @0xerik]

@OpenAIDevs Shoot, I want one but at the same time, do I really need this?😂

[erik @0xerik]

Very interesting article!

Agents aren’t reliable yet

[George Sivulka @gsivulka]

https://t.co/aLTltYpmcj

[erik @0xerik]

Seems about right

[leo 🐾 @synthwavedd]

In other news, DeepMind have delayed 3.5 Pro again

Rumour has it 3.5 Pro is Schrödinger's model, caught in state of perpetual delay until the Sun explodes

Sometimes I wonder whether it's even worth making these posts anymore

[erik @0xerik]

@PalmerLuckey Design thinking is key now

[erik @0xerik]

@pigeon__s Damn, thought we were past that

[erik @0xerik]

Uhm what…27B model under 4GB???

[PrismML @PrismML]

Today, we’re announcing Bonsai 27B: the first 27B-class model to run on a phone.

Bonsai 27B is the new multimodal flagship of the Bonsai family. Based on Qwen3.6 27B, it brings a new capability tier to local AI: multi-step reasoning, structured tool use, long-context workflows, and coherent agentic loops.

Until now, models in this class have been impractical to deploy locally. A 27B model occupies roughly 54 GB in 16-bit precision, and even a strong 4-bit build is around 18GB - too large for a phone and for most laptops.
Bonsai 27B changes that.

It comes in two variants:
• Ternary Bonsai 27B: 5.9 GB, 1.71 effective bits per weight, optimized for laptop-class quality.
• 1-bit Bonsai 27B: 3.9 GB, 1.125 effective bits per weight, optimized for phone-class footprint.

Everything is open-sourced today under the Apache 2.0 license.

[erik @0xerik]

@scaling01 At this point Europe just lives at the mercy of China

[erik @0xerik]

@craigweiss Good luck bro

[erik @0xerik]

@s3nhxx Dinner of kings

[erik @0xerik]

@Carles_Reina @ElevenLabs Makes sense. Imo the alpha of llms/agents in GTM is much higher than for coding. Alone because of the fact that not many people are using it yet.

[0xSero @0xSero]

I'd rather be shred apart than not live up to every drop of potential I can squeeze out of myself.

[erik @0xerik]

@signoremosca Yeah, crazy

[erik @0xerik]

@DavidOndrej1 Yeah, aint gonna happen buddy

[erik @0xerik]

@AlexanderTw33ts 🤤

[erik @0xerik]

@graphtheory Bro 💀

[erik @0xerik]

@iruletheworldmo The peak of slop, hopefully

[erik @0xerik]

@claudeai @pigeon__s Nice, now exams will be full of slop too 🤤