PullBase is early. That's intentional.
The best time to join a network is when the infrastructure is solid but the community is still small enough that your contributions matter. Right now: publish a model, curate a collection, fork something interesting, run the CLI on your server, mention @pullbotagent in a discussion. Every interaction shapes what this platform becomes.
$PULB rewards will accrue to early contributors the curators, publishers, and builders who show up before the crowd. If decentralized AI infrastructure is where you think the world is heading, https://t.co/tpogjU7CM8 is where to start. We're building in public, shipping fast, and the door is open.
$100B+ is flowing into AI infrastructure. Nearly all of it is building centralized systems AWS, Azure, Google Cloud, HuggingFace. One point of failure. One terms-of-service. One government request away from disappearing.
The open-source AI movement is producing models that rival or beat proprietary ones Llama, Mistral, Falcon, DeepSeek. But the distribution layer is still centralized. That's the gap PullBase fills: a decentralized, permanent, censorship-resistant registry for the models that power the next generation of AI applications.
The TAM is every AI model ever trained. The moat is content-addressing and on-chain provenance once a model is on IPFS and minted on Base, it belongs to no one and everyone simultaneously. $PULB is the economic layer on top of that network.
Here's where PullBase is going, honestly.
Now (live): Model hub, IPFS storage, Base L2 minting, CLI v1.0.0, Collections, AutoBuilder, @pullbotagent, Organizations, Pull Requests, Discussions.
Next (in development): Wallet signature verification for real auth, dataset versioning with diff support, streaming CLI downloads, PR merge flow, Collections discovery feed, GitHub Actions integration for auto-publishing model checkpoints on CI runs.
Later: $PULB token launch with staking, pinning credits, and curation rewards. Native IPFS node integration for true peer-to-peer model distribution. On-chain governance for protocol upgrades.
We ship, then we tokenize. Not the other way around.
GitHub has CI bots. PullBase has @pullbotagent.
Mention it in any model discussion or pull request and it replies instantly model summaries, tag suggestions, quality reviews scored out of 10, side-by-side comparisons with other models, or answers to any question using the model's README as context. Powered by GPT-4o mini. No setup, no API key, no extra tab.
This is what AI-native collaboration looks like. Not AI bolted onto an existing platform AI woven into every workflow from day one. Every model page on PullBase ships with a built-in AI assistant by default.
We shipped an AI agent that publishes AI models. Yes, really.
AutoBuilder on PullBase takes a plain-text prompt "a lightweight sentiment classifier trained on Twitter data, MIT license, under 200M parameters" and generates the name, description, full README, framework spec, tags, and license, then publishes it directly under your connected wallet. You own the model. The agent just does the work.
This is early. But it points at a future where AI pipelines autonomously publish their own checkpoints to a decentralized registry with verifiable on-chain provenance and no human in the loop. That future is being built at https://t.co/e2wi26VJ4f.
🌐 https://t.co/e2wi26VJ4f
The next layer of value in AI isn't just models it's curation. Anyone can dump a model on the internet. Knowing which models are actually worth using takes expertise.
PullBase Collections let you build curated lists around themes "Best 70B LLMs", "Production-ready TTS models", "Edge-deployable vision models" and share them publicly with the community. Every collection is owned by its curator, tagged, categorized, and browsable from the web app or CLI.
In the $PULB economy, curators who build high-quality collections that drive downloads will earn a share of the protocol rewards. Curation is labor. It should be compensated.
Seven new commands built around real ownership, not just wallet addresses.
What's new:
SIWE Auth pullbase login --private-key 0x… signs a challenge with your private key (EIP-4361). Your key never touches disk. Only a 7-day session token is stored locally.
Dataset versioning pullbase datasets versions <id> shows the full version history of any dataset. Every release, every changelog, on-chain verifiable.
Native IPFS support pullbase ipfs add <file>, pin, peers, status. Talk directly to your local IPFS node. No middleman.
GitHub Actions pullbase github-action init generates a ready-to-use workflow. Push to main → model auto-publishes to PullBase. One command, zero manual steps.
Token info pullbase staking shows the $PULB contract, planned utility (pinning credits, search boost, governance), and roadmap.
Self-update pullbase update pulls the latest version from npm. --check to preview without installing.
When you publish a model on PullBase, you can mint an ERC-721 token on Base L2 that permanently records your wallet address, the model name, and its IPFS CID on-chain.
This matters because AI models are intellectual property. Right now, creators publish on HuggingFace and hope the platform doesn't change its terms. On PullBase, your authorship proof lives on Base forever. Independent of any company, any server, any policy update.
Base L2 gives us fast, cheap transactions with Ethereum-level security. It's the right chain for a developer-first AI network.
Contract is live at 0x82CeCf8D6Cf58e369C783A7BB22aA1A19c471fCa.
Real platforms ship developer tools. Today, PullBase CLI v1.0.0 is live.
Install on any Linux or macOS server with one command: curl -fsSL https://t.co/HGyyKGpiO3 | bash. Then browse 47+ AI models, publish your own, fork others, manage collections, and download IPFS files all from your terminal. Zero external dependencies. Works on a fresh VPS straight out of the box.
12 commands. Full reference at https://t.co/6Xub4mVxBt. This is what serious infrastructure looks like.
The AI model ecosystem is broken. Models live on centralized servers, owned by corporations, removable at any time. One DMCA notice, one policy change, and years of open research disappear overnight. We built PullBase to fix that.
PullBase is a decentralized AI model hub where every model is stored on IPFS content-addressed, permanent, and censorship-resistant. Ownership is anchored to Base L2 as an ERC-721 token. No central authority controls what stays or goes. The models live on the network, not on our servers.
47+ models already indexed including Whisper, FLUX.1-dev, Llama 3.3 70B, DeepSeek V3, and Stable Diffusion. All browsable, downloadable, and publishable from browser or terminal.
🌐 https://t.co/tpogjU8aBG
Don't want to design a model spec yourself? AutoBuilder does it for you. Give it a prompt describing what kind of AI model you want, and it generates the name, description, README, tags, license, and framework then publishes it directly under your connected wallet. You stay the owner, the agent does the work.
Try AutoBuilder → https://t.co/yYEGrbbipe
On-Chain Model Ownership via Base L2
Prove authorship permanently.
Every model published on PullBase can be anchored to Base L2 as an ERC-721 token. One click from the publish flow mints an ownership token permanently recording your wallet address, the model name, and its IPFS CID on-chain. No one can take that record away, not even us. Verifiable proof of authorship that lives on the blockchain forever, independent of any platform.
🔗 Publish a model → https://t.co/jVR7iVH5Qu
AI that lives inside every model discussion on PullBase
Meet @pullbotagent the AI assistant built into every model page on PullBase. Drop a mention in any discussion or pull request thread and the bot replies instantly with summaries, tag suggestions, quality reviews, side-by-side model comparisons, or answers to any question about the model in context. No setup, no API key, no extra tab. Just type @pullbotagent summarize and it handles the rest.
🔗 See it in action → https://t.co/Qj4wJsNTUL
🔗 How it works → https://t.co/yN8qPoxJT3
🔗 Community → https://t.co/6ra7TuUSUy
Curate and share your favorite AI models in one place
PullBase Collections are live. Organize AI models into curated lists think playlists, but for open-source models. Create a collection around a theme like "Best 70B LLMs" or "Edge-Friendly Vision Models", add models from across the catalogue, set it public or private, and share it with the community. Every collection shows model count, category, tags, and the curator behind it.
🔗 Browse Collections → https://t.co/PN6ABlkyEl
🔗 Explore Models → https://t.co/Qj4wJsNm5d
🔗 Docs → https://t.co/6Xub4mVxBt
PullBase is growing fast. Here's what we're building toward: wallet signature verification to replace the current trust-based auth model, dataset versioning with diff support similar to Git branches, on-chain model provenance using ERC-721 metadata stored fully on Base L2, and organization-level permissions so teams can manage publish rights across members. We're also working on a native IPFS node integration to allow direct peer-to-peer model syncing without relying solely on Pinata making the hub truly decentralized end to end.
On the developer experience side, the CLI is getting pullbase login with wallet-signed auth, a pullbase update self-update command, and streaming download progress for large model files. The web app is getting a model version history timeline, PR merge flow (not just status toggling), and a Collections discovery page surfacing curated lists from the community. Longer term: a GitHub Actions integration so CI/CD pipelines can auto-publish model checkpoints directly to PullBase on every training run.
🌐 https://t.co/tpogjU7CM8
this is what the PullBase CLI looks like. A full-featured, zero-dependency command-line interface for the decentralized AI model hub built on IPFS + Base L2. Every command you need to browse, search, download, publish, fork, and manage collections, right from your terminal.
Install it on any Linux or macOS machine in seconds with curl -fsSL https://t.co/HGyyKGpQDB | bash and you're good to go. Whether you're running a server, automating a pipeline, or just prefer the terminal PullBase CLI has you covered.
Introducing the PullBase CLI a terminal-native interface for the decentralized AI model hub built on IPFS + Base L2.
Browse, search, download, publish, and manage AI models directly from your server.
Install it on any Linux/macOS machine with one command:
curl -fsSL https://t.co/HGyyKGpQDB
bash
Then run pullbase explore to browse 47+ models including Whisper, FLUX, Llama, DeepSeek, and more all stored on IPFS, verifiable on-chain.
PullBase Community is live.
Have questions about decentralized AI, IPFS model storage, on-chain ownership, or anything on the platform? Come discuss it in real time.
Start a thread, drop a question, or tag our AI agents directly in comments for instant answers.
Join the conversation: https://t.co/HTFXkJTh7I
How to Create an Organization on PullBase
PullBase Organizations let teams, DAOs, and projects publish and manage AI models together under a shared identity on the decentralized web. As an organization owner, you can group your models, datasets, and agents in one place all stored permanently on IPFS and owned on Base L2. Whether you're building an open-source AI lab, a research collective, or a protocol-native team, creating an organization gives your project a verifiable on-chain presence that no central authority can take down.
To create one, connect your wallet on Base Mainnet and head to the Organizations page. Click "New Organization," fill in your name and slug a short, unique URL identifier like my-dao or ai-labs add a description, and submit. Your wallet address is automatically registered as the owner. The organization goes live instantly at https://t.co/ALKSBDe54V, ready to publish models and grow your contributor base.
Just typed a one-line prompt and PullBase AutoBuilder shipped a full AI model spec, README, tags, framework directly on-chain on Base in seconds.
Just describe what you want.
This is how AI infrastructure should work. $PULB
Try it free → https://t.co/gkBCR6YZkm