{"id":"opengradient","title":"OpenGradient","content":"**OpenGradient** is a technology company focused on developing decentralized infrastructure for artificial intelligence, aiming to merge blockchain technology with AI. [\\[15\\]](#cite-id-7HRO31bO334VR5kf). The project's stated mission is to create an open, verifiable, and user-owned AI ecosystem that empowers the permissionless creation, distribution, and deployment of AI models and applications. This is intended to counteract the \"black box\" nature of centralized AI platforms and to democratize model ownership. [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW) [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp). The company's public-facing history shows several iterations, with initial concepts focusing on a decentralized AI model hub and later developments centered on a proprietary Layer 1 blockchain designed for verifiable AI computation and persistent memory. [\\[2\\]](#cite-id-pY4JDy4iBhJDQDpe) [\\[7\\]](#cite-id-90VLONZBBk7B9Aqa).\n\n[YOUTUBE@VID](https://youtube.com/watch?v=KD1BMaNLuts)\n\n## Overview\n\nOpenGradient aims to address the limitations of centralized AI systems, which often lack transparency, control user data, and operate as opaque services. Its goal is to prevent the \"data fracking\" common with large technology companies by ensuring user data and models remain owned by the user. [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp). It is an end-to-end decentralized infrastructure network designed for AI model hosting, secure execution, and application deployment. By integrating blockchain technology, the project seeks to establish an open standard for AI computation and data management, making AI inference and data processing accessible directly from smart contracts. [\\[7\\]](#cite-id-90VLONZBBk7B9Aqa) [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW). The company's philosophy is rooted in open-source principles, decentralization, and user empowerment, with a research-first approach that prioritizes security, privacy, and user data ownership. [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW).\n\nThe project has evolved through different phases. Early reports in 2024 described OpenGradient as a decentralized AI model hub built on networks like Bittensor for compute and [Arweave](https://iq.wiki/wiki/arweave) for storage, positioning it as a direct competitor to platforms like [Hugging Face](https://iq.wiki/wiki/hugging-face). [\\[2\\]](#cite-id-pY4JDy4iBhJDQDpe). Later announcements in 2025 shifted the focus to the development of a foundational Layer 1 blockchain, termed \"The L1 Network for Open Intelligence.\" This iteration emphasizes building \"AI that remembers\" through proprietary technologies for persistent memory, verifiable computation, and user-owned data, enabling AI systems to learn and evolve while preserving user autonomy. [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW) [\\[3\\]](#cite-id-mNyIj1vZBdUgDzMo).\n\nThe articulated goal is to build the foundational layer for secure and transparent AI systems that serve human agency rather than corporate interests. The platform and its tools are designed for developers to build and deploy \"sovereign agents,\" where users retain ownership of their data and models, and all computational processes are verifiable on-chain. [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp).\n\n[YOUTUBE@VID](https://youtube.com/watch?v=dDXTOPRKyjg)\n\n​\n\n## History and Funding\n\nOn August 19, 2025, OpenGradient publicly announced it had raised $8.5 million to develop its decentralized AI infrastructure, framing the project as a Layer 1 blockchain. Venture capital investors included [a16z crypto](https://iq.wiki/wiki/a16z-crypto), [Coinbase](https://iq.wiki/wiki/coinbase) Ventures, SV Angel, and Foresight Ventures. Strategic investors included [Celestia](https://iq.wiki/wiki/celestia) and NEAR, with notable angel investors [Balaji Srinivasan](https://iq.wiki/wiki/balaji-srinivasan), [Illia Polosukhin](https://iq.wiki/wiki/illia-polosukhin) (co-founder of [NEAR](https://iq.wiki/wiki/near-protocol)), and [Sandeep Nailwal](https://iq.wiki/wiki/sandeep-nailwal) (co-founder of [Polygon](https://iq.wiki/wiki/polygon)). [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW).\n\nA subsequent announcement on September 23, 2025, for the launch of its MemSync product, stated the company had raised a total of $9.5 million. This announcement listed the new leadership as co-founders [Matthew Wang](https://iq.wiki/wiki/matthew-wang) (CEO) and Adam Balogh (CTO), who are described as veterans from Google, Meta, and Palantir. Investors mentioned in this release were [a16z crypto](https://iq.wiki/wiki/a16z-crypto), SVA (Struck Ventures), and SALT. [\\[8\\]](#cite-id-vWtdl6ebR6uMlYVp).\n\n## Technology\n\nOpenGradient's technology has evolved from an early concept of a decentralized model hub to a comprehensive Layer 1 blockchain ecosystem for AI.\n\n### OpenGradient Network Architecture\n\nThe current iteration of OpenGradient is a foundational Layer 1 blockchain built specifically for AI. It is an EVM-compatible network that utilizes a proprietary [Hybrid](https://iq.wiki/wiki/hybrid) AI Compute Architecture (HACA). HACA is designed to scale and secure on-chain AI workflows by using node specialization, dedicating different types of nodes to specific tasks such as inference, agentic reasoning, and statistical analysis to achieve efficiency and scalability while integrating decentralized GPUs and specialized accelerators. [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp) [\\[15\\]](#cite-id-7HRO31bO334VR5kf). It is designed to provide high-performance, secure, and confidential infrastructure for on-chain AI activities. The network is built on a specialized architecture composed of several node types: Full Nodes (The Judge), Inference Nodes (The Sprinter), Storage Nodes (The Librarian), and Data Nodes (The Scout). This structure is designed to securely run the entire AI workflow on-chain, from data access and pre-processing to inference computation, with the blockchain used to settle and attribute every inference. [\\[7\\]](#cite-id-90VLONZBBk7B9Aqa) [\\[9\\]](#cite-id-zGOoPB1XUUVOHcOG) [\\[12\\]](#cite-id-I7jQ5rD7f4jIhJ2c). The network aims to make every AI agent toolcall, model inference, and API request verifiable on-chain using cryptographic proofs. [\\[6\\]](#cite-id-hYRy6xer7Gmgpy0A) [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp).\n\n### Nova Testnet\n\nThe OpenGradient Nova [Testnet](https://iq.wiki/wiki/testnet) launched on October 1, 2025, initiating what the project termed the \"Third Era of Blockspace,\" where intelligence becomes a native, verifiable component of a ledger. The [testnet](https://iq.wiki/wiki/testnet) embeds AI computation and its proof directly into the consensus mechanism to solve issues of latency, opacity, and cost associated with using off-chain AI in blockchain applications. [\\[12\\]](#cite-id-I7jQ5rD7f4jIhJ2c).\n\nIts core is a decentralized AI execution layer that integrates AI inference directly into the block production process. Key components include a Parallelised Inference Pre-Execution Engine (PIPE) to prevent slow AI models from delaying block production and an Inference Data Availability (DA) layer where cryptographic proofs of computation are included in the block data for independent verification. The Neuro Stack is a framework that allows development teams to build their own layer-2 rollups with custom tokens while using OpenGradient's AI computation layer as a shared service, effectively providing \"AI-as-a-service\" for the modular blockchain ecosystem. [\\[12\\]](#cite-id-I7jQ5rD7f4jIhJ2c).\n\n### Verifiable On-Chain AI\n\nA core feature of the network is enabling AI computations to be executed and cryptographically verified on-chain. This is achieved through a **Verifiable Inference SDK** and a network of secure hardware enclaves to prove that a specific model ran with certain inputs, ensuring the integrity of the output without revealing confidential data. The platform also focuses on **Data Provenance**, a system for tracking the origin and lineage of data used in the inference process to ensure transparency. [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW) [\\[9\\]](#cite-id-zGOoPB1XUUVOHcOG). The platform offers developers a choice between Zero-Knowledge (ZK) proofs for mathematical guarantees and Trusted Execution Environment (TEE) attestations for hardware-based security. [\\[12\\]](#cite-id-I7jQ5rD7f4jIhJ2c).\n\nThe platform also integrates Zero-Knowledge Machine Learning (ZKML), partnering with projects like EZKL and [Lagrange](https://iq.wiki/wiki/lagrange) Labs. This allows AI model computations to be verified using SNARKs ([Succinct](https://iq.wiki/wiki/succinct) Non-Interactive Arguments of Knowledge), providing trust-minimized inference. [\\[6\\]](#cite-id-hYRy6xer7Gmgpy0A).\n\n## Core Products and Features\n\n### MemSync\n\nLaunched on September 23, 2025, **MemSync** is a universal memory layer for AI assistants like ChatGPT, Claude, and Perplexity. It is designed to solve the problem of \"context loss\" by creating a persistent, secure, and unified memory system that carries user context across different platforms, applications, and devices. [\\[8\\]](#cite-id-vWtdl6ebR6uMlYVp). It gives users granular control over their data, which is kept in an encrypted, on-device \"memory vault.\" [\\[6\\]](#cite-id-hYRy6xer7Gmgpy0A) [\\[1\\]](#cite-id-gYmVBcHjboUnWTNW).\n\nAccording to a press release, internal benchmarks by OpenGradient showed MemSync achieved a 243% improvement in memory retrieval and response quality over OpenAI's \"industry standard\" solution. [\\[8\\]](#cite-id-vWtdl6ebR6uMlYVp). In separate benchmarks replicating the Locomo test, the company reported that MemSync outperformed competitors by at least 18.9%, with particular strength in retaining details across multiple conversations. [\\[11\\]](#cite-id-rkPGx85IPPduqG9Q). The long-term vision for the technology includes the creation of \"digital twins\"—AI representations of individuals built from their public and authorized private data, with early demonstrations including twins of public figures like Naval Ravikant and Sydney Sweeney. MemSync launched with a free tier, a Chrome extension, and a developer API. [\\[8\\]](#cite-id-vWtdl6ebR6uMlYVp).\n\n#### Architecture\n\nOn September 15, 2025, OpenGradient detailed the architecture of MemSync, which is built on three pillars inspired by human psychology. [\\[11\\]](#cite-id-rkPGx85IPPduqG9Q).\n\n* **Memory Collection:** This pillar uses a dual-memory system. **Semantic Memories** are stable, long-term facts about a user (e.g., identity, core interests), while **Episodic Memories** are temporary, time-specific details about current events or projects. A \"Smart Consolidation\" process manages this information through four operations: Create, Update, Reinforce (based on an importance score), and Delete.\n* **Memory Formation:** The system categorizes memories (e.g., `Identity`, `Career`, `Health`) and synthesizes them into evolving, high-level summaries called \"profiles.\" These profiles provide concise snapshots of a user's personality to maintain long-term context.\n* **Memory Retrieval:** A three-stage process surfaces relevant information. It begins with a broad **Vector Search** for semantically related memories. A **Contextual Precision Ranking** model then re-ranks the results for direct relevance to the current conversation. Finally, an **Optimization Layer** blends the top-ranked semantic and episodic memories to provide the AI model with rich, layered context.\n\n### BitQuant\n\nBitQuant is an open-source, crypto-native AI trading agent developed by OpenGradient using 'not-a-dev' tooling. It was open-sourced under an MIT license on October 29, 2025, following a private beta phase involving over 50,000 users. The agent is designed to function as an AI-native quantitative framework that interprets natural language commands and converts them into verifiable on-chain transactions. [\\[13\\]](#cite-id-cPsZwElkQcN0XMA2) [\\[14\\]](#cite-id-3VsnmjnV4tt4PLQi).\n\nBitQuant's modular architecture is composed of three main components: an oracle, a brain, and a trader. [\\[13\\]](#cite-id-cPsZwElkQcN0XMA2).\n\n* **Oracle:** The data and execution layer, which aggregates information from on-chain sources like the [Solana](https://iq.wiki/wiki/solana) RPC and DeFi protocols such as Orca and Solend, as well as from data aggregators like [CoinGecko](https://iq.wiki/wiki/coingecko) and DeFiLlama.\n* **Brain:** The decision-making engine, which uses a Router Large Language Model (LLM) to interpret user prompts and direct them to specialist agents for analytics or investment execution.\n* **Trader:** The execution component, which allows users to perform actions like \"hedge my SOL exposure\" through natural language. The agent then guides the user through the transaction process.\n\nThe framework is built to be extensible, allowing developers to build custom agents. It also integrates Bittensor hooks to enable decentralized AI computation. Its features include DeFi analytics, portfolio management, and a natural language interface for interacting with complex on-chain data. [\\[13\\]](#cite-id-cPsZwElkQcN0XMA2) [\\[14\\]](#cite-id-3VsnmjnV4tt4PLQi).\n\n### OpenGradient Model Hub\n\nThe **OpenGradient Model Hub** is a central product that acts as a decentralized registry for AI models. It is positioned as a censorship-resistant and community-owned alternative to centralized platforms like [Hugging Face](https://iq.wiki/wiki/hugging-face). [\\[5\\]](#cite-id-JQdp7EWPYWcvQftN) [\\[2\\]](#cite-id-pY4JDy4iBhJDQDpe). In its current form, the Model Hub is a web application frontend (`hub.opengradient.ai`) built on its decentralized storage partner, **Walrus**. It allows users to permissionlessly upload, manage, and version models of various architectures (e.g., neural networks, LLMs) for use on the OpenGradient network. Access is provided via the web UI and a Python SDK for more advanced management. On December 19, 2025, the team announced that the Model Hub had surpassed 1,000 live, verifiable models hosted on its testnet. [\\[7\\]](#cite-id-90VLONZBBk7B9Aqa) [\\[6\\]](#cite-id-hYRy6xer7Gmgpy0A) [\\[19\\]](#cite-id-ZrIQebqlzim13IGW).\n\n### Twins (twin.fun)\n\nTwins is a platform built on OpenGradient's infrastructure, designed around the concept of creating and interacting with AI \"twins.\" The platform has its own smart contracts, data access tools, and a roadmap, with defined roles for \"Creators\" and \"Traders.\" It functions as a distinct application layer within the OpenGradient ecosystem. [\\[9\\]](#cite-id-zGOoPB1XUUVOHcOG) [\\[7\\]](#cite-id-90VLONZBBk7B9Aqa). The concept is powered by the MemSync architecture to create AI representations of individuals from their public and authorized private data. Use cases include interacting with digital versions of experts and historical figures or creating personal AI assistants to automate tasks. [\\[8\\]](#cite-id-vWtdl6ebR6uMlYVp) [\\[12\\]](#cite-id-I7jQ5rD7f4jIhJ2c).\n\n## Events\n\n### Model-thon (Jan-Feb 2025)\n\nIn January and February 2025, OpenGradient held its inaugural Model-thon, a competitive event co-sponsored by [Allora](https://iq.wiki/wiki/allora). The event was designed to encourage developers to create and deploy high-performance machine learning models on the OpenGradient platform, with a focus on [Web3](https://iq.wiki/wiki/web3) applications. Participants submitted models in ONNX format to the OpenGradient Model Hub to compete across several tracks. [\\[16\\]](#cite-id-fJYtCU0IWF1hlERn).\n\nThe event featured three main tracks: a BTC Spot Forecast Track, an ETH Spot Forecast Track, and a Freestyle Track. The two spot forecast tracks required participants to build models that could predict hourly price returns for BTC/USDT and ETH/USDT, respectively. The Freestyle track allowed for more creative submissions judged on originality and usefulness to the [Web3](https://iq.wiki/wiki/web3) ecosystem. Performance in the forecast tracks was measured using a custom metric called Mean Z-tanh Absolute Error (MZTAE), which was designed to better reward accurate predictions of extreme price movements. Following the competition period, winners in each track were awarded prizes. [\\[16\\]](#cite-id-fJYtCU0IWF1hlERn).\n\n## Developer Tools\n\nOpenGradient provides a suite of tools for building AI-powered applications. [\\[10\\]](#cite-id-nXpAvjvx31gNF8UM).\n\n* **OpenGradient SDK:** A Python SDK that enables both on-chain and off-chain applications to leverage the network's infrastructure for model management and decentralized, verified AI inference. [\\[10\\]](#cite-id-nXpAvjvx31gNF8UM) [\\[9\\]](#cite-id-zGOoPB1XUUVOHcOG).\n* **SolidML:** A [Solidity](https://iq.wiki/wiki/solidity) framework for enabling machine learning model execution directly within EVM-compatible smart contracts. It allows for complex logic on-chain, including model inference, data preprocessing, and interaction with price feeds. [\\[10\\]](#cite-id-nXpAvjvx31gNF8UM) [\\[4\\]](#cite-id-sqbx5ZfQJlBWjuVp).\n* **Agent Stack:** A technology stack for creating decentralized and verifiable AI agents that are executed on-chain. [\\[10\\]](#cite-id-nXpAvjvx31gNF8UM).\n* **Neuro Stack:** An open-source development stack for building AI-native Layer 2 blockchains, referred to as \"Neuro-Chains.\" This framework allows developers to create sovereign and verifiable AI applications or agents that run entirely on-chain, settling inference transactions on the main OpenGradient network. Built on the HACA infrastructure, the Neuro Stack provides customizable blockspace and seamless access to secure inference methods like ZKML and TEE without requiring complex setup. It is designed to foster a modular and scalable ecosystem through permissionless composability. The first project built using the stack is an AI-native blockchain for [DePIN](https://iq.wiki/wiki/depin) developed in partnership with Peri Labs. [\\[17\\]](#cite-id-0WR7SXgtbtnZi2lp).\n* **AlphaSense:** A tool designed for developers to create \"verifiable AI workflows\" that provide AI agents with trusted and powerful data signals. [\\[9\\]](#cite-id-zGOoPB1XUUVOHcOG).\n* **LangChain Integration:** The platform integrates with the LangChain framework via the `Langchain-Opengradient` package. This allows developers to build AI agents that leverage specialized, verifiable machine learning models from OpenGradient's network. The primary component, the `OpenGradientToolkit`, enables an agent to use complex ML models as tools. This integration is designed to avoid context window pollution by having the model process data on the OpenGradient network and return only the final result to the agent. A key feature is the ability to produce verifiable outputs; inferences are secured with ZKML or TEEs, and the transaction's execution trace is recorded and verified on the OpenGradient blockchain for trustless computation. [\\[18\\]](#cite-id-ir2Dhb9lEDoIAETH).","summary":"OpenGradient is a tech company creating a Layer 1 blockchain for open intelligence. It offers a decentralized model hub for hosting, sharing, and running AI models with on-chain verifiable inference and user-owned data to rival centralized platforms.","images":[{"id":"QmYpZHZqJMkWJTCsEJ5mvkdLD7f1otsAkFFYtcnaLYzoq4","type":"image/jpeg, image/png"}],"categories":[{"id":"dapps","title":"dapps"}],"tags":[{"id":"AI"},{"id":"Protocols"},{"id":"Organizations"}],"media":[{"id":"https://www.youtube.com/watch?v=KD1BMaNLuts","name":"KD1BMaNLuts","caption":"","thumbnail":"https://www.youtube.com/watch?v=KD1BMaNLuts","source":"YOUTUBE"},{"id":"https://www.youtube.com/watch?v=dDXTOPRKyjg","name":"dDXTOPRKyjg","caption":"","thumbnail":"https://www.youtube.com/watch?v=dDXTOPRKyjg","source":"YOUTUBE"}],"metadata":[{"id":"website","value":"https://www.opengradient.ai/"},{"id":"twitter_profile","value":"https://x.com/OpenGradient"},{"id":"linkedin_profile","value":"https://www.linkedin.com/company/opengradientlabs"},{"id":"github_profile","value":"https://github.com/OpenGradient"},{"id":"discord_profile","value":"https://discord.gg/2t5sx5BCpB"},{"id":"medium_profile","value":"https://opengradient.medium.com/"},{"id":"references","value":"[{\"id\":\"gYmVBcHjboUnWTNW\",\"url\":\"https://www.opengradient.ai/\",\"description\":\"Official website overview\",\"timestamp\":1766417771015},{\"id\":\"pY4JDy4iBhJDQDpe\",\"url\":\"https://www.theblock.co/post/322677/opengradient-launches-decentralized-ai-model-hub-challenging-traditional-platforms\",\"description\":\"The Block article on OpenGradient's launch\",\"timestamp\":1766417771015},{\"id\":\"mNyIj1vZBdUgDzMo\",\"url\":\"https://x.com/OpenGradient\",\"description\":\"OpenGradient X profile bio\",\"timestamp\":1766417771015},{\"id\":\"sqbx5ZfQJlBWjuVp\",\"url\":\"https://www.opengradient.ai/blog/opengradient-raises-8-5m-to-decentralize-ai\",\"description\":\"Details on sovereign agents and on-chain verification\",\"timestamp\":1766417771015},{\"id\":\"JQdp7EWPYWcvQftN\",\"url\":\"https://cryptorank.io/ico/opengradient\",\"description\":\"CryptoRank data on OpenGradient funding\",\"timestamp\":1766417771015},{\"id\":\"hYRy6xer7Gmgpy0A\",\"url\":\"https://www.linkedin.com/company/opengradientlabs\",\"description\":\"LinkedIn company profile and funding details\",\"timestamp\":1766417771015},{\"id\":\"90VLONZBBk7B9Aqa\",\"url\":\"https://docs.opengradient.ai/about/\",\"description\":\"OpenGradient official documentation overview\",\"timestamp\":1766420033021},{\"id\":\"vWtdl6ebR6uMlYVp\",\"url\":\"https://www.prnewswire.com/news-releases/opengradient-launches-memsync-universal-memory-layer-for-ai-assistants-302560572.html\",\"description\":\"Press release on MemSync launch and funding\",\"timestamp\":1766420033021},{\"id\":\"zGOoPB1XUUVOHcOG\",\"url\":\"https://docs.opengradient.ai/learn/\",\"description\":\"OpenGradient architecture details\",\"timestamp\":1766420033021},{\"id\":\"nXpAvjvx31gNF8UM\",\"url\":\"https://docs.opengradient.ai/developers/\",\"description\":\"Developer Tools Overview\",\"timestamp\":1766420033021},{\"id\":\"rkPGx85IPPduqG9Q\",\"url\":\"https://www.opengradient.ai/blog/building-better-ai-memory-the-architecture-behind-memsync\",\"description\":\"OpenGradient blog post on MemSync architecture and benchmarks\",\"timestamp\":1766420592953},{\"id\":\"I7jQ5rD7f4jIhJ2c\",\"url\":\"https://www.opengradient.ai/blog/introducing-opengradient-nova-testnet\",\"description\":\"Introducing the OpenGradient Nova Testnet\",\"timestamp\":1766421286987},{\"id\":\"cPsZwElkQcN0XMA2\",\"url\":\"https://www.opengradient.ai/blog/open-sourcing-bitquant\",\"description\":\"Open-Sourcing BitQuant blog post\",\"timestamp\":1766421286987},{\"id\":\"3VsnmjnV4tt4PLQi\",\"url\":\"https://www.opengradient.ai/blog/meet-bitquant-crypto-ai-quant\",\"description\":\"Meet BitQuant blog post\",\"timestamp\":1766421286987},{\"id\":\"7HRO31bO334VR5kf\",\"url\":\"https://www.opengradient.ai/blog/introducing-opengradient\",\"description\":\"Introductory blog post on OpenGradient\",\"timestamp\":1766424420979},{\"id\":\"fJYtCU0IWF1hlERn\",\"url\":\"https://docs.opengradient.ai/modelthon/\",\"description\":\"OpenGradient Modelthon Detailed Information\",\"timestamp\":1766424420979},{\"id\":\"0WR7SXgtbtnZi2lp\",\"url\":\"https://www.opengradient.ai/blog/introducing-the-opengradient-neuro-stack\",\"description\":\"Introducing the OpenGradient Neuro Stack\",\"timestamp\":1766424420979},{\"id\":\"ir2Dhb9lEDoIAETH\",\"url\":\"https://www.opengradient.ai/blog/opengradient-langchain-integration\",\"description\":\"OpenGradient LangChain Integration blog\",\"timestamp\":1766424420979},{\"id\":\"ZrIQebqlzim13IGW\",\"url\":\"https://x.com/OpenGradient/status/2002026332373573654\",\"description\":\"OpenGradient announces 1,000+ models on Model Hub\",\"timestamp\":1766425887538}]"},{"id":"previous_cid","value":"\"https://ipfs.everipedia.org/ipfs/QmXWxiEKFGRy49DkFRMbpHNXPNhxJGe8kFKtJb6hwJasCA\""},{"id":"commit-message","value":"\"Removed introduction and overview sections\""},{"id":"previous_cid","value":"QmXWxiEKFGRy49DkFRMbpHNXPNhxJGe8kFKtJb6hwJasCA"}],"events":[{"id":"a5ccc96a-fe06-4333-a9a4-325b031d8ebd","date":"2024-01","title":"OpenGradient Founded","type":"CREATED","description":"OpenGradient was founded to build a decentralized infrastructure for artificial intelligence.","link":null,"multiDateStart":null,"multiDateEnd":null,"continent":null,"country":null},{"id":"915f3222-c815-47f6-923c-0db3c40e65fa","date":"2024-10","title":"Announced $8.5M Seed Funding Round","type":"DEFAULT","description":"OpenGradient announced it raised an $8.5 million seed funding round with participation from a16z crypto, Coinbase Ventures, and others.","link":"https://www.opengradient.ai/blog/opengradient-raises-8-5m-to-decentralize-ai","multiDateStart":null,"multiDateEnd":null,"continent":null,"country":null},{"id":"a4bd2576-0cb2-47fe-b8a7-5a86b6051a92","date":"2024-12","title":"Launched OpenGradient Model Hub","type":"DEFAULT","description":"The public version of the flagship product, the OpenGradient Model Hub, was launched as a decentralized platform for hosting and running AI models.","link":"https://www.opengradient.ai/blog/introducing-the-opengradient-model-hub","multiDateStart":null,"multiDateEnd":null,"continent":null,"country":null},{"id":"d7b62d6b-fb1d-4af2-9f0c-f33c6ffa67a9","date":"2025-07","title":"Introduced MemSync Technology","type":"DEFAULT","description":"OpenGradient introduced MemSync, its proprietary technology for creating portable and persistent memory for AI models.","link":"https://www.opengradient.ai/blog/introducing-memsync","multiDateStart":null,"multiDateEnd":null,"continent":null,"country":null}],"user":{"id":"0x8af7a19a26d8fbc48defb35aefb15ec8c407f889"},"author":{"id":"0x8af7a19a26d8fbc48defb35aefb15ec8c407f889"},"operator":{"id":"0x1E23b34d3106F0C1c74D17f2Cd0F65cdb039b138"},"language":"en","version":1,"linkedWikis":{"blockchains":[],"founders":["matthew-wang"],"speakers":[]},"recentActivity":"{\"items\":[{\"id\":\"990a3d27-d8c5-4da2-8b7f-61fe82690c75\",\"title\":\"OpenGradient\",\"description\":\"OpenGradient is a tech company creating a Layer 1 blockchain for open intelligence. It offers a decentralized model hub for hosting, sharing, and running AI models with on-chain verifiable inference and user-owned data to rival centralized platforms.\",\"timestamp\":\"2025-12-22T17:56:26.769Z\",\"category\":\"dapps\",\"status\":{\"icon\":\"RiGlobalLine\",\"label\":\"Wiki Updated\",\"iconClassName\":\"text-green-500\"},\"user\":{\"name\":\"0x8af7a19a26d8fbc48defb35aefb15ec8c407f889\",\"address\":\"0x1E23b34d3106F0C1c74D17f2Cd0F65cdb039b138\"},\"button\":{\"label\":\"View Summary\",\"icon\":\"RiFileTextLine\"},\"summarySections\":[{\"title\":\"Content\",\"subtitle\":\"Added videos, internal links, and an update about the model hub.\",\"variant\":\"modified\",\"changeCount\":4,\"changes\":[\"Added an embedded YouTube video to the introduction.\",\"Added an embedded YouTube video to the 'Overview' section.\",\"In the 'History and Funding' section, added internal links for Balaji Srinivasan, Illia Polosukhin, and NEAR.\",\"In the 'OpenGradient Model Hub' section, added an announcement that the hub had surpassed 1,000 live, verifiable models [[19]](#cite-id-ZrIQebqlzim13IGW).\"]},{\"title\":\"Videos\",\"subtitle\":\"Added two YouTube videos.\",\"variant\":\"added\",\"changeCount\":2,\"changes\":[\"Added video: https://www.youtube.com/watch?v=KD1BMaNLuts\",\"Added video: https://www.youtube.com/watch?v=dDXTOPRKyjg\"]},{\"title\":\"Images\",\"subtitle\":\"Removed four images from the gallery.\",\"variant\":\"removed\",\"changeCount\":4,\"changes\":[\"Removed image with ID: QmYJhZx3u6kWPHEG3UzvHXyiJT7E9U1JUyYzLJDsJb9eE4\",\"Removed image with ID: Qmdo5fYpgnFPNVKUhGBMBuJ8b7ehBjKTgbapHvAHbs4S1E\",\"Removed image with ID: QmcLQbsVJjWhMMB1RyA1LET4ugzn49mLchYsY8EEPuwYVQ\",\"Removed image with ID: QmUFhfc39jMUUR187SxbD83vxc2eCXN4eiyFtapDNts9zQ\"]},{\"title\":\"References\",\"subtitle\":\"A new reference was added for a project milestone.\",\"variant\":\"modified\",\"changeCount\":1,\"changes\":[\"Added reference: OpenGradient announces 1,000+ models on Model Hub [[19]](#cite-id-ZrIQebqlzim13IGW).\"]},{\"title\":\"Linked Wikis\",\"subtitle\":\"Added 3 new links to related wiki pages.\",\"variant\":\"added\",\"changeCount\":3,\"changes\":[\"Added link to 'balaji-srinivasan'\",\"Added link to 'illia-polosukhin'\",\"Added link to 'near-protocol'\"]}]}]}"}