Hellas

serverless AI secured by web3 -

[Hellas @hellasdotai]

Great episode with our founders @statusfailed and @0xBaltar. Take a listen!

[RAFA Finance @RAFA_AI]

AI Podcast w/ Hellas Network @hellasdotai

An ALL-IQ discussion about deterministic AI, decentralised compute, #agenticAI, and the PAIN POINTS being solved by their network and our @RAFA_AI platform.

Watch. Learn, and walk away smarter in under an hour.

This one's for bookmarking and sharing.

Got questions? Let us know in the comments 👇

[Hellas @hellasdotai]

AI hallucinations are not just a model problem.

They are also an infrastructure problem.

If inference becomes cheap enough, users can run outputs across multiple models, prompts, and checks instead of trusting one answer blindly.

Hellas makes that future more realistic.

[George @0xBaltar]

RDMA for everyone!*

2x Strix Halo 128GB
+
2x Aliexpress USB4 Cables
+
custom Linux kernel module + userspace
=
~236 usable GB VRAM
~40Gbps full duplex bandwidth, single-digit μs latency

versus ethernet:

11x faster finetuning
14 -> 20tok/s tg at batch=1 on MiniMax-M2.7-AWQ

[Hellas @hellasdotai]

@system_monarch It's like handing someone the best ship ever built. The experienced captain goes further every time, because the ship still doesn't sail itself.

[Hellas @hellasdotai]

@ThePeterMick Where's the vibe vs craft line for you? we just let AI write an entire kernel module turning consumer Thunderbolt into InfiniBand.

https://t.co/eLPW7WKNoi

[Hellas @hellasdotai]

@geeksforgeeks Double-edged sword in action, we let AI write this entire kernel module. https://t.co/eLPW7WKNoi

[Hellas @hellasdotai]

Determinism changes what AI can become.

If the same input can reliably produce the same output, AI moves from “probabilistic black box” toward an infrastructure that developers can actually build on.

That is one of the core ideas behind Hellas.

[Hellas @hellasdotai]

So here is hellas in one pitch :)

Hellas is building a permissionless network for tensor compute. Developers get access to compute without relying on a single centralised provider.

GPU owners can make idle hardware productive.

Models become easier to serve. AI becomes more open, more liquid, and less dependent on closed infrastructure.

[Hellas @hellasdotai]

@swyx Do they make the compute market more liquid? And give users more compiler customisation?

[Hellas @hellasdotai]

@charles_irl *Verifiable Inference-Maxxing

[Hellas @hellasdotai]

@AlgodTrading For the more curious minds, we are building a verifiable compute network. https://t.co/HQbnRjrN5U

[Hellas @hellasdotai]

https://t.co/LhYRUVERPE

[Hellas @hellasdotai]

@antirez Have you compared against a reference implementation to rule out a subtle quantisation or GGUF bug?

[Hellas @hellasdotai]

It's been one week since centralized AI hit the cost wall. Two days 'til they realize compute should've been liquid all along. -Dr. Manhattan

[Hellas @hellasdotai]

@weswinder The issue is executives were sold on the idea of historic productivity gains while almost no one actually has the know-how, agentic orchestration, scaffolding, harness design, to bring about this productivity.

[Hellas @hellasdotai]

AI should not be controlled by a small group of companies with the power to decide who gets compute, what model they get, and when access gets throttled.

Hellas is building the market layer for open AI compute.

More providers.
More models.
More choice.

[Hellas @hellasdotai]

@JordanSchachtel UBI will create a demoralised and frustrated population. Better to embrace the future and start learning how to customise your inference calls and orchestrate agents.

[Hellas @hellasdotai]

@JordanSchachtel Which human qualities will have instrinct value in this new age?

[Hellas @hellasdotai]

@fabianstelzer We said AI was being built like a cathedral. We did not expect them to take it this literally.

[Hellas @hellasdotai]

@const_reborn Precisely, AI is being built like a Cathedral filled with dogma and supersitiion. The worlds needs to be able to opt out and have a permissionless bazaar.