1:11:13

💎 ⚔️

[Ansem 🐂🀄️ @blknoiz06]

@0xLurixn @ThePumponomics bullshit

[leyten @leyten]

Big update for Compute Network.

The X suspension for @c0mputeAI after our rebrand and the appeal is still pending, so I'm posting this from my personal account.

I've been working behind the scenes to upgrade both our native and browser worker models, and to attract retail consumers to the network.

Everything below is live today:

1. The network now runs one model: Qwen3.8 27B Uncensored. Native multimodal, tools, web search, vision, thinking.

2. Introducing plans:
- Free: 20 credits per day.
- Pro: $12 per month, 300 per day.
- Max: $30 per month, 750 per day.

A typical message costs about 1 credit. Paid in USDC, Stripe card checkout coming. This will make Compute Network more accessible to retail consumers.

https://t.co/LWk2b6Aj8Y

3. Billing moved to per token. You pay for what the model actually generates, and failed jobs refund automatically.

4. The browser worker upgraded from Qwen3 8B to Qwen3.5 Uncensored, selected automatically for your GPU: 9B on stronger cards, 4B on 8GB cards.

5. Native workers run one command and serve the same model as everyone: 16GB+ NVIDIA, 24GB+ AMD, or 32GB+ Apple Silicon, now on Metal.

If you're currently still running the old worker:
npx @compute-network/worker@latest

6. The economics favor suppliers: a GPU that stays online earns more than a subscription costs.

7. Reliability work across the stack: long thinking answers no longer time out, image generation from chat delivers every time, and the data page shows exact numbers.

Private, uncensored and decentralized AI made possible by https://t.co/J8hceFLyG4.

[1:11:13 @1eleventhirteen]

When you feel the shortness of breath, try exhaling first, before then taking a quick deep inhale, and a loooooong slow exhale. And try breathing from your belly more, rather than your chest. Diaphragmatic breath.
Trust that your body knows how to breathe. You got this.

Source: Am therapist.

[1:11:13 @1eleventhirteen]

@marvCalledIt @Yakuzamotosan @li_mina_l I’m not sure it’s over.

[1:11:13 @1eleventhirteen]

@li_mina_l @MoonsAndLambos Hallways all day

[1:11:13 @1eleventhirteen]

@yunalspy Lol yes, I guess figurative arms will have to do for now.

[leyten @leyten]

1/ We ran Kimi K3 (@Kimi_Moonshot), a 2.78 trillion parameter model, on 16 distinct RTX 5090s and 6 distinct RTX Pro 6000s spread across Europe over the open internet.

Two rings, tested back to back:
• 16 RTX 5090s: about 2.2 tok/s
• 6 RTX Pro 6000s: about 7.3 tok/s

[leyten @leyten]

We need to take control of AI in our own hands, this is unacceptable.

Everyday I get remembered why I'm building @c0mputeAI.

[Julian @orithellama]

So, many asked, is @ICPXProtocol a competitor to @c0mputeAI?

The simple answer is no.

ICPX does offer GPUs and compute as well, but the main product is running one-click deploys and jobs.

Where we could leverage c0mputeAI as a service provider of decentralised compute.

Why PVP if we can collab?

@leyten what we say mate?

[Chris Gilbert @0xgilbert]

UsePod × c0mpute: private, uncensored inference is live

We’ve integrated c0mpute, a decentralized inference network, into the UsePod marketplace. If you want models that don’t refuse, don’t store your prompts, and run on a distributed worker network, there’s a model for that!

Find them under the new Uncensored filter on the marketplace. Same drop-in OpenAI-compatible API, same USDC balance, just choose one of the new models.

[「 𝕲𝖔𝖔𝖓」 @goon_crypto]

Every AI company you're paying is reading your prompts.
Most people accept that as the cost of access.

@c0mputeAI is a network running right now where the GPU processing your request has no idea who sent it.

OpenAI, Anthropic, Google all log your conversations. They can revoke your access tomorrow. Their models refuse entire categories of questions based on editorial decisions you had no vote on.

c0mpute is a decentralised inference network. Regular people share GPU power. You get private, uncensored AI.

The flow:
1. You send a message
2. An orchestrator finds an available worker based on your tier
3. The worker runs the model and streams tokens back
4. Job finishes. Orchestrator discards it. No conversation database. Only token counts kept for billing.

Two tiers:
➜ Pro: Qwen3 8B Uncensored, runs in your browser via WebGPU, 10 credits per message
➜ Max: Qwen3.5 27B or SuperGemma4 26B, native GPU workers, web search + vision, 15 credits

The worker gets your prompt text and nothing else. No name. No wallet. No account. They can't know it's you.

Workers earn 70% of every job in USDC, 80% if they stake $ZERO. The orchestrator weights selection by tokens/sec, so faster workers earn more but the whole pool gets work.

When OpenAI goes down, millions lose access. When they update their refusal list, topics that worked last month stop working. No central authority means no single point to pressure into blocking content.

You gave up your prompts, your trust, and your options for convenient access.

Are you running uncensored inference for anything real yet, or still routing everything through filtered endpoints?

$ZERO is live on @virtuals_io now!

[tonyGewrit🟦 @tonyGewrit]

its like buying $ray in october 2023 at 40m mcap. the token then run to 2bn.

a simple but necessary infrastructure project going up in value NOT because of hype but user demand.

today, AI inference is what needs to be “used”, its the necessary infrastructure.

the projects at the frontier: $ZERO $SQUIRE $ICPX.

big numbers but the opportunity is that big; AI.

[1:11:13 @1eleventhirteen]

@tonyGewrit 💯

[leyten @leyten]

Demo for the sharded inference engine on: https://t.co/wNR4oxM8vp

gpt-oss-120b on 4090's, max input context 95k, max output context infinite (tok/s degrades with context)

queue of max 12 prompts concurrently, use it wisely

[1:11:13 @1eleventhirteen]

@shahh @NVIDIAAI It pumped because of the tech that inspired engagement from @NVIDIAAI

@c0mputeAI is on to something big

[Ethan Kravitz @EthosVentures]

This a great writeup from Modal on why speculative decoding (specdec) is the future of open source inference and IMO this is extremely bullish for @c0mputeAI solana:EmcxFTNVDqyLHp11NvwvLZ4D7LKGbG9i7B8RF7dwpump

speculative decoding turns token prediction from a serial process to a parallel process. c0mpute's Shard engine distributes this parallel process across nodes in the network, rapidly splitting enough work to overcome latency

after renting my own servers and trying the proof of concept myself, I am convinced there is still a ton of room for optimization. there is low hanging fruit like constantly updated geographic optimization, but there are also huge wins from improved drafter models.

c0mpute's GLM 5.2 PoC uses a GLM 4 9B draft model but if this is upgraded to a native GLM 5 MTP sidecar or a custom trained DFlash model like this article describes, we could see improvements of 100% or more. imagine what breakthroughs will come once Fable is back online.

the future of decentralized inference is bright and specdec will lead the way! highly recommend reading the blog below