Qubic (@_Qubic_) header
Qubic (@_Qubic_) profile picture

No VCs. No premine. No fees. Just pure decentralized intelligence. Bare-metal L1 · UPoW-powered AI · 15.5M TPS · Aigarth building AGI. Power to the network.

Qubic
Joined May 2023
Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)Photo by Qubic (@_Qubic_)
Pinned Tweet
Most mining throws away trillions of guesses. Qubic’s doesn’t. With Ant Colony Mining, miners build on each other’s best results, like ants following a trail. Progress is kept, week after week. Live on mainnet now.
28
160
17
643
49,401
When Qubic's science team started on ARC-AGI-3, its best score was 0.13. The final competition submission reached 0.41. No GPUs. No LLMs. Small brains evolved on ordinary CPUs.
11
82
8
373
12,329
In absolute terms it's still a low score, and the team says so. The goal isn't to memorize answers. It's to learn new rules as efficiently as a human does. Join Today's Open Science AMA at 3:00 PM UTC to hear directly from our Scientists.
2
6
73
1,719
A smart contract only knows what's on its own chain. Oracle Machines change that. Live on $Qubic since February, they bring outside data on-chain, answering a query in about seven network ticks.
7
76
3
373
10,282
$Qubic Open Science AMA is back! Qubic's science team, @VivancosDavid and @josesanchezhb, are taking your questions on ARC-AGI-3 and what the team is working on next. 📅 Thursday, 19 November | 3 PM UTC | 10 AM EST 📍Live on X. Drop your questions below 👇
2
61
5
318
11,009
At Qubic AGI Lab in Basel, @josesanchezhb made the case for a different path to AGI: grow intelligence instead of training it. Today's AI is trained once in a data centre, then frozen. Neuraxon is built to keep learning while it acts.
8
82
4
378
13,310
Most blockchains make you choose between speed and cost. Qubic doesn't. Transfers are feeless and final instantly. Smart contracts in C++ run directly on the hardware, with no virtual machine slowing them down. The result: 15.5 million transactions per second on mainnet, verified by CertiK.
7
89
13
375
18,019
Most AI is judged on its final score. Neuraxon 3.0 will be judged on how fast it learns. The next brain design from Qubic’s science team scores each brain on how quickly it reduces its mistakes on games it has never seen. That rewards real learning, not lucky guesses. 🧵
9
90
7
390
12,600
"Not a mind, a monopoly." At Qubic AGI Lab in Basel, @josesanchezhb named the real risk of AI. It isn't a machine that thinks for itself. It's one organisation that builds it, tests it and decides who can use it. His answer is to decentralize it. Highlights from the evening 👇
8
71
5
342
13,763
Community Space #3 recap is live 🎙️ @PjdubluLLY walked us through @Qubic_Capital (QCAP): a community DAO that pools funds into shares of Qubic smart contracts, with every decision made by holder vote. From how QVAULT governance works to why no single admin controls the treasury, it was a deep dive into one of Qubic’s most active communities.
4
50
3
283
11,023
Who controls the network you rely on? On Qubic, no single party does. That is by design. Here is how the network keeps control spread out 🧵
11
82
6
386
16,166
Underneath it all, the code is open source for anyone to review. Nodes run directly on hardware with no operating system. The mining that earns Computors their place also contributes to AI training. Your keys. Your network. Your AI.
1
16
1
95
3,263