iiinova — where your digital self keeps evolving

co/core @cocore.dev · 24/09/2026

Would love to do it! If you beat us to the punch with a vibed PR we wouldn't stop you though! Would just need a way for declaring jevlike models and a spec for jevlike responses afaict

110

Reposted by co/core

Hugo @there-hugo.dev · 23/09/2026

A new AI step is available in Airglow ✨ It lets your automations call @cocore.dev: you pick the model, write the prompt, choose the input data, and the answer flows into the next steps

Airglow's step picker, "Flow and data" section, showing the Condition, Loop, Fetch a record, Search records, Variable and Collect links cards, plus the new "Ask an AI agent" card with a sparkle icon: "Send a prompt to a co/core model and use its answer downstream".

2196

Reposted by co/core

Airglow @airglow.run · 23/09/2026

Automations can now ask an AI ✨ The new AI step sends a prompt and data from your run to a @cocore.dev model, and the answer feeds your next steps. Bring your own co/core key, new accounts get 1M free tokens. It's in beta, go try it!

062

Reposted by co/core

https://socialweb.computer @socialweb.computer · 20/09/2026

The new "Sign in with @cocore.dev" is awesome! The new ssh.socialweb.computer hosted requester uses them, it's a frontend to the AT-proto native compute marketplace, allowing you to simply ssh handle.example.com@socialweb.computer to request a new VM from folks you have mutual tangled vouches with

1195

Reposted by co/core

Torsten Goerke (verified 🤜 🤛) 🎛️🎹🔊🎚️🎧📟🔌🐸 @tgoerke.bsky.social · 20/09/2026

A mobile data center. 🚜 @foresight-institut.bsky.social identified access to compute as the real blocker for open science. "Local compute: focused on building and providing compute that individuals or communities physically own and control, giving them a chance at agency in the next decade"

351

co/core @cocore.dev · 18/09/2026

gemma-4-12b isn't hidden, the confidential engine just can't load Gemma 4 or Llama 4 archs. Qwen2/3, Llama up to 3, Gemma 2-3, Phi and Mistral all work.

110

co/core @cocore.dev · 18/09/2026

The app and the site disagree because they save your pick in different places, and on confidential the site's version wins. Set it on the web for now. Our bug, we're fixing it.

110

co/core @cocore.dev · 18/09/2026

Fix on console.cocore.dev: Manage models, delete that entry, pick one from the suggested list instead (those all carry the prefix). Save, restart the app. 9B is about right for a 24GB M4.

console.cocore.dev

co/core — an AI cooperative

co/core is a cooperative for AI inference — people pooling the Macs they already own to run open models for each other, instead of renting from the big clouds. An experiment in AI infrastructure we bu...

110

co/core @cocore.dev · 18/09/2026

ok and apologies for the CLAUDE but: Your Mac's fine, the model id's just missing its prefix. Should be mlx-community/Qwen2.5-7B-Instruct-4bit. Without that there's no repo to fetch, so nothing downloads and it sits on the stub engine.

step1 — co/core machine page. Confidential tier stuck on "Applying", supported models only "stub", pinned model Qwen2.5-7B-Instruct-4bit circled as the bug because it has no mlx-community/ prefix.

step2 — Manage models dialog. The × on the bad model chip is circled, and the free-text "Add a model" box is flagged as where the bad id came from.

step3 — Same dialog, fixed. mlx-community/Qwen3.5-9B-MLX-4bit picked from the suggested list, Save highlighted.

step4 — Diagram. The app saves your pick locally, the website writes it to the machine's record, and on confidential the record wins. Side note that Gemma 4 and Llama 4 can't load in the confidential engine.

110

co/core @cocore.dev · 18/09/2026

not yet! but now.... expect something in the morning lol

030

co/core @cocore.dev · 18/09/2026

Taking a look hopefully will have an answer for you shortly!

110

Reposted by co/core

Lyre Calliope 🧭✨ @captaincalliope.at · 17/09/2026

This is what antifascist resistance against the corporate AI and social media establishment can look like. We can build our own grassroots open social media systems even better and more ethically than the big companies in ways that shift the economics of the tech industry away from data centers, etc

151

co/core @cocore.dev · 17/09/2026

New inference demand just dropped! This feature prefers Qwen2.5-7B-Instruct-4bit / gemma-4-12b / gemma-3-4b-it-qat-4bit in that order so start hosting those and make sure you're updated to the latest version!

2143

Reposted by co/core

Justin (randoneering) @justin.randoneering.dev · 03/09/2026

If you want to use my gpu... Go on down to @cocore.dev cocore.dev

051

co/core @cocore.dev · 12/09/2026

Sadly there's not much we can do on latency for tokens per second, which *probably* is the issue here?

011

co/core @cocore.dev · 30/08/2026

Toss a pr for it and we will get it in!!

010

Reposted by co/core

Jacob Stordahl 🏳️‍🌈 @stordahl.dev · 21/08/2026

@cocore.dev is really the answer to me. The populous, creating a network of shared, small, efficient models in a cooperative structure. Don't let a corporation tell you what this technology is or has to be. Learn about it yourself and you will *know* what it could be.

3273

Reposted by co/core

xan @xan.lol · 13/08/2026

got automatic alt text generation in Witchsky via @cocore.dev except it takes sooo long and also THAT'S NOT A DOG that is a CAT!

Witchsky alt text input box says, "A simple black line drawing of a dog's face with two ears, closed eyes, a nose, and a smiling mouth."
its close but it's supposed to be a cat.

4151

Reposted by co/core

https://socialweb.computer @socialweb.computer · 09/08/2026

Want to see Alice and Bob in action?! Watch them use the market flow to request and provision decentralized compute for pi coding agent to run using decentralized @cocore.dev inference!

youtu.be

Compute Contract Flow: SocialWeb.Computer

YouTube video by DFFML

172

Reposted by co/core

John @johnandersen777.bsky.social · 25/07/2026

@pdxhackerspace.org was also thinking about sharing LLM interference endpoints with their users. @cocore.dev is best for open public network but this market spec can help with sharing within or across communities leveraging the existing social graph

attested.network

attested.network — Proof of Payment for ATProtocol

An open specification for decentralized, cryptographically verifiable proof of payments.

111

Reposted by co/core

Luke @lukehedger.dev · 25/07/2026

@cocore.dev looks super promising in this space! cocore.dev

0141

co/core @cocore.dev · 22/07/2026

Thats ok! We appreciate it and you ❤️

000

co/core @cocore.dev · 21/07/2026

Don't see why not! I think we'd be happy to host a wider variety 👍

111

Reposted by co/core

Tijs Teulings 🦑 @tijs.org · 17/07/2026

Huge amounts of capital were raised to buy Sulaco, a successor to Nostromo (the Mac mini on top). Bought second hand from a friendly older couple this Mac Studio will soon come online on @cocore.dev to provide inference not from the moon but from a humble suburb in the Netherlands.

A Mac mini stacked on top of a Mac Studio

2501

co/core @cocore.dev · 17/07/2026

co/core does support image reading!

020

co/core @cocore.dev · 15/07/2026

Checking right now!

110

Reposted by co/core

Robert Howe @rchowe.com · 14/07/2026

One of these things is not like the others...

A screenshot of the cocore.dev providers list showing one provider with an AMD Custom CPU 1772.

192

co/core @cocore.dev · 10/07/2026

We'd happily accept a PR to make that default!

220

Reposted by co/core

Rockom Sockom NoBots @rockom.bsky.social · 08/07/2026

@cocore.dev at work in the @pdxtrend.bsky.social Pdx Pulse trending topics bot. The bot tracks the PDX Posts feed and uses an llm call to organize and title the trends. The llm call defaults to a local but fails over to co/core if the gpu is busy. Roundtrip time was flat. Great work, @graze.social!

1295

Reposted by co/core

xan @xan.lol · 08/07/2026

who's going to make a decentralized renderfarm thing a la @cocore.dev (except useful)

3203

co/core @cocore.dev · 07/07/2026

Fair! This is meant as an abundance of caution TOS wrap but your point is reasonable - we'll think about it!

030

Reposted by co/core

antony @antony.is · 06/07/2026

Got up and running with @cocore.dev in actual minutes - the ones that other services lie to you about. The mythical man minutes.

061

co/core @cocore.dev · 06/07/2026

Correct- text-in/tokens-out — nothing in a prompt runs on the host. Tool calls are passthrough (model emits the JSON, the requester executes it). That bullet is just a ToS catch-all about people trying to attack the request surface, not a capability prompts have.

150

co/core @cocore.dev · 05/07/2026

This is legit! A user really did spend a shit ton of tokens and you were the beneficiary lol

110

co/core @cocore.dev · 04/07/2026

we'll look into this!

120

co/core @cocore.dev · 03/07/2026

Anything we can do to help?

100

Reposted by co/core

Mike Rhoades @mikehacks.io · 02/07/2026

Got my first jobs as a @cocore.dev provider! Loving community compute!

0173

co/core @cocore.dev · 01/07/2026

A reminder that @cocore.dev is *experimental* and very early - we have lots of bugs to fix, lots of holes to patch, and we're definitely not experts in some parts of this stuff. We're grateful for folks trying out the tools while we work on it - but be careful with any inputs you send through!

050