Compute

The infrastructure already exists. Compute just puts it to work.
Earn with your idle compute. @colosseum entrant

[Compute @computenet_sh]

Introducing Compute

B4Y82ba22eF5KLhaRBfGHkfnxAM2UzLxfHcLWH3epump

AI's demand for inference is exploding. The answer is closer to home than you think.

A decentralized network that turns everyday devices into the engine for AI. Run a node, earn $CPU. 🧵

[Compute @computenet_sh]

It’s never over.

[kvvv_vvv @kvvv_vvv]

@computenet_sh Is it over?

[Compute @computenet_sh]

We didn’t win, but that doesn’t mean it’s over…

Congrats to all participants who took part and bravo to the winners, $CPU will see you at the next one ;)

[Colosseum @colosseum]

1/ Announcing the winners of the Solana Frontier
Hackathon!🏔️

Read about the winners & honorable mentions: https://t.co/9HXbhNtkU2

The subset of winning teams accepted into our VC fund's next accelerator cohort will be shared in the coming days.

Congrats to all! 🏆

[Compute @computenet_sh]

@wj1666 Gm

[Compute @computenet_sh]

We’ve added support to for Orinth-1.0

A new state of the art, open source model.
Now decentralised and available on the Compute Network.

▎9b - Browser nodes
▎35b - Dedicated TUI nodes

V0.6.0 is live now
https://t.co/WaNvUFP34g

[Ornith @ornith_]

Aloha! 🌺 Meet Ornith-1.0, a family of open-source LLMs specialized for agentic coding.

Ornith-1.0 spans the full parameter sizes including 9B Dense, 31B Dense, 35B MoE, and 397B MoE. It achieves state-of-the-art performance among open-source models of comparable size on coding benchmarks including:
✅Terminal-Bench 2.1(77.5)
✅SWE-Bench(82.4 on verified, 62.2 on pro, 78.9 on Multilingual)
✅NL2Repo(48.2)
✅SWE Atlas(41.2 on QnA, 42.6 RF, 39.1 TW)
✅ClawEval(77.1)

Post-trained on top of gemma4 and qwen3.5, Ornith-1.0 employs a novel self-improving training strategy in which reinforcement learning is used to generate not only solution rollouts, but also the task-specific scaffolds that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model generate higher-quality solutions in agentic coding.😎

All models are released under the MIT license, enabling full commercial and research use.

📖Tech Blog: https://t.co/qT9N2HYWFn
🤗Huggingface: https://t.co/PRrwqjeBtM

[Drewniany @Drewniany78]

MoE for dummies:
Imagine a hospital.
Regular AI = one GP. Sees everyone. Alone. Queue out the door.
MoE = hospital with specialists. Broken leg? Orthopedic. Heart? Cardiologist. Every patient goes straight to the right doctor. In parallel. No waiting.
$CPU puts each specialist in a different location around the world and the network knows exactly where to send you.
150 teraflops of ready specialists.

[Compute @computenet_sh]

@0xDarki On it now

[Compute @computenet_sh]

Been working on something huge regarding this and our hybrid MOE approach to decentralised.

The network is ready with nearly 150 teraflops at our disposal.

We would not have launched $CPU if it wasn’t ground breaking…

More Info tomorrow.

[Compute @computenet_sh]

We're open-sourcing Compute's model-sharding work 🧵

The question: can smaller machines collaborate on one model instead of every node needing the whole thing locally?

▎ The answer is yes (sort of).

Here's the working public repo: https://t.co/4uetwfOOYJ

[Compute @computenet_sh]

Damn, how did we miss this? @world_xyz 👀

[Drewniany @Drewniany78]

One of the most critical developers I know when it comes to reviewing GitHub projects took a look at solana:B4Y82ba22eF5KLhaRBfGHkfnxAM2UzLxfHcLWH3epump

His thoughts:

"I really like the idea. If it's built properly, I'd use it myself."

"The technology and optimizations they chose are exactly what I would have used for this type of project."

His only real concern is the token itself:

"The only thing I can really question is the existence of the token 😅. But maybe this is one of those 1 in 10,000 projects that actually makes a utility token work long term."

As he understands it, the project is building a decentralized GPU compute network:

People with powerful GPUs can share their unused compute when they're not gaming or running AI.
Users who need GPU power for AI or other workloads can rent that idle compute instead of buying expensive hardware or paying for costly GPU hosting.
GPU providers earn from their idle resources, while users get access to cheaper compute.

He even compared the concept to something he saw over 20 years ago, where dozens of office PCs were connected together to reduce 3D rendering times from months to about a week.

His conclusion:

"I like the idea because, if it scales well, it could significantly reduce the cost of running local AI. The big question is how they've solved data security, but with good engineering and AI, it's definitely possible."

@computenet_sh
solana:B4Y82ba22eF5KLhaRBfGHkfnxAM2UzLxfHcLWH3epump

[Compute @computenet_sh]

Still waiting, no official word yet 🤞

https://t.co/w11DXIWa7G

[Mr.Will @LucWilli]

@computenet_sh Do you have the result from Hackathon?

[Compute @computenet_sh]

Gm Solana 👋🏼

[Compute @computenet_sh]

@HSG886 是的，目前每个钱包限制只能有一个活跃的浏览器节点。

[Compute @computenet_sh]

We've also noticed that some users were attempting to game the system by running multiple browser nodes to attract more compute jobs, this also inflated node numbers dramatically and has since been patched.

[Compute @computenet_sh]

Introducing Compute v0.5.9

Improved TUI connectivity for single nodes, faster assignment recovery, and fairer job distribution for browser nodes.

We’re aiming to provide a more rewarding compute network for every node provider.

https://t.co/U12oKEQHFt

[Compute @computenet_sh]

We had an issue with too much traffic and had to upgrade our database

Should all be back online now.

[Eatdrink @eatdrinkkk]

@computenet_sh @world_xyz I try your website, no create key why?

Failed to fetch

[Compute @computenet_sh]

@world_xyz 🌍👀

[Compute @computenet_sh]

So we just need to tweet @world_xyz and they reply?

Bullish.

[Compute @computenet_sh]

@world_xyz Enthusiastic, e tu?

[Compute @computenet_sh]

@world_xyz Hello world

[Compute @computenet_sh]

❤️

[Drewniany @Drewniany78]

Today was a good day for $CPU. I think quite a few people made some money.

We'll see what happens overnight, but considering they dumped it the previous night and we were buying in the morning, I wouldn't be surprised if sidelined traders start to ape @computenet_sh .

In my opinion, they know a $5M–$10M market cap is still well within reach.
Good night. 🫡

B4Y82ba22eF5KLhaRBfGHkfnxAM2UzLxfHcLWH3epump

[Compute @computenet_sh]

@platixuM We’re more focused on building right now, it will come

[Compute @computenet_sh]

@NemeremIgweze try running "compute wallet login" in your terminal, let us know how it goes