The First Web3 LLM

Hi, it’s Marc.

Welcome to +329 new readers since our last update. Today, we cover:

• ASI-1 Mini
• Sentient Chat
• AI Agents
• Top reads
• Other highlights

Let’s jump in.

?? Subscribe to 51x - AI x Crypto

The Ultimate AI Agent Intelligence Platform

Discover, compare, and track AI vendors in real time — powered by research and data, not hype. Join the waitlist for exclusive early access ??

ASI-1 Mini: The First Web3 LLM

Fetch AI has introduced ASI-1 Mini, the world’s first Web3-native large language model, purpose-built for complex agentic workflows.

The goal? To deliver industry-leading performance at a fraction of the cost with decentralization.

ASI-1 Mini is setting a new standard by using advanced Knowledge Graph technology.

• Smart Data Maps: Think of Knowledge Graphs as dynamic maps that chart how data points connect. They enable AI to understand relationships, track context, and remember details.
• Personalized Interaction: By capturing your preferences and past interactions, these graphs allow ASI1: Mini to deliver tailored, context-rich responses—evolving with your needs over time.

How It Works

• Picks the Right Tool: ASI-1 Mini uses a “Mixture of Models” approach, selecting the best AI for each job—like grabbing the perfect wrench from a toolbox.
• Team Effort: Its “Mixture of Agents” means multiple AI helpers work together, tackling complex tasks faster and smarter.
• Low Cost, High Power: It runs on just two GPUs, not a room full of pricey hardware, making it affordable for more people to use.

AI can feel like a mystery box: you get an answer, but how? ASI-1 Mini changes that. It offers clearer glimpses into its decision-making.

What Sets ASI-1 Mini Apart:

• Cost-Efficient Power: Running on just two GPUs, ASI-1 Mini offers top-tier performance without heavy hardware investments.
• Adaptive Reasoning: With four dynamic reasoning modes—Multi-Step, Complete, Optimized, and Short—this model tailors its thought process to the task, balancing depth and speed.
• Scalable Architecture: By blending specialized models and autonomous agents in a three-layered system, it’s built to tackle everything from intricate data analysis to enterprise-level decision-making.
• Expanding Capabilities: Soon to support context windows up to 10M tokens, making it ideal for analyzing lengthy documents or large-scale datasets.

ASI-1 Mini is available as a tiered freemium product for $FET token holders.

“ASI-1 Mini is the first major product from the ASI Alliance’s innovation stack, marking the beginning of the ASI:<Train/> rollout and a new era of community-owned AI. This launch sets the foundation for a decentralized ecosystem where the Web3 community can invest, train, and directly benefit from cutting-edge AI models”

— Humayun Sheikh, CEO of Fetch.ai

?? Subscribe to 51x - AI x Crypto

ASI:<Train/>

ASI:<Train/> is a platform where everyday people can help create and own advanced AI models, working together as a community. It lets everyone contribute and share in the benefits, making AI development more open and fair.

Developed by Artificial Superintelligence Alliance, ASI-1 Mini isn’t the best model on the market just yet, it’s truly exciting to witness the rapid progress in Web3-native LLM development.

What’s next: They will release more features like enhanced agentic tool-calling and multi-modal capabilities. It’ll:

• Make split-second decisions in real time.
• Run tasks without constant human nudges.
• Handle text, images, and more with ease.

More links on :

• ASI1 Documentation
• Artificial Superintelligence (ASI) Alliance Vision Paper
• Artificial Superintelligence Alliance Documentation
• A Taste of the Superintelligence Vision Paper by Ben Goertzel

New Report: State of AI in 2025 – Beyond The Hype

Read Our Latest Report

Sentient Chat: The Open AGI Play

Sentient AGI’s new platform, Sentient Chat, integrates over 20 specialized AI agents—offering real-time web browsing and autonomous workflows. Positioned as a competitor to Perplexity, the platform aims to expand its agent ecosystem through community contributions.

Why it matters: Backed by an $85M seed round (including Peter Thiel’s Founders Fund), Sentient AGI is positioning itself as the open alternative to centralized AI ecosystems. Instead of a closed system, Sentient Chat invites the community to contribute, ensuring faster iteration, broader expertise, and reduced reliance on corporate AI silos.

Zooming in: There are 4 agents available in the agent hub:

1. Messari Copilot: It leverages Messari's extensive crypto research and datasets to provide accurate, context-rich answers to complex cryptocurrency questions.
2. Open RSS3: It enables rapid creation of AI agents, leveraging the Open Web for verifiable computing, which supports complex operations like DeFi and DeSci with precision and reliability.
3. QnA3: It is an AI agent by QnA3 AI designed for crypto traders, using real-time data to provide tailored investment advice.
4. KITE: Developed by Kite AI, it focuses on analyzing Ethereum transactions to detect potential fraud, using advanced AI to evaluate and summarize transaction legitimacy.

Waitlist: waitlist-chat.sentient.xyz

?? Subscribe to 51x - AI x Crypto

Top AI Agents by mindshare

Some agents you want to keep an eye on:

• $VIRTUAL ($477M): A framework and infrastructure designed to create, deploy, and monetize autonomous AI agents across various platforms and applications, such as gaming, entertainment, social interactions, and more.
• $AIXBT ($269M): A crypto-savvy AI that scans markets and social media to deliver trading insights and alpha opportunities.
• $TRAC ($182M): A decentralized knowledge-sharing protocol and blockchain-based platform that uses a Decentralized Knowledge Graph (DKG) to organize, verify, and share trusted data and knowledge assets across industries.
• $FAI ( $145M): AI built with publicly available large language models (LLMs) and operates on the Base blockchain.
• $ARC($108M): A modular toolkit for creating AI bots that are highly customizable, especially for crypto ecosystems.
• $PAAL ($105M): Personalized, advanced AI assistants and applications, particularly for crypto investors and broader digital communities.
• $GRIFFAIN ($53M): Creates AI tools that make AI agents smarter and more efficient, like enabling them to trade on DeFi protocols.
• $SWARMS ($54M): Focuses on teams of AI bots (“swarms”) that work together to tackle big challenges.
• $ZEREBRO ($33M): Autonomous AI system designed to create, distribute, and analyze content across decentralized and social platforms like X and Instagram.

Top Quotes

• “lol there’s a new frontier AI model dropping every week now. China back again with QWEN by Alibaba, which rivals DeepSeek’s flagship R1. Oh & it’s open source. It’s also 5% the size of DeepSeek’s flagship R1 model (32B vs. 671B) which means it can fit on your laptop & just over the boundary of fitting on your phone (h/t Hoseeb). Seems they achieved this through distillation + scaled reinforcement learning. China really beating the U.S. at the open source game.” — Ejaaz Ahamadeen
• “Architecture is destiny.” — Chris Dixon
• “Instead of chasing full autonomy, we want multimodal systems that work with people—enhancing creativity, decision-making, and problem-solving together.” — Maja

Top Reads

• OpenAI x L1 Playbook. Link
• The AI Betting Genius That's Outsmarting EVERYONE [Podcast]. Link
• An Integrated Approach to the ASI Ecosystem. Link
• Decentralizing AI with GensynAI. Link
• Crypto x AI: Investing Across The Stack. Link
• How to Create Crypto AI Agents With ElizaOS & Gelato. Link
• AI16Z + ELIZA: The Blueprint for Web3’s Next AI-Driven Leap. Link

Other Alpha:

• Olas introduced RateRover, an autonomous AI agent on Arbitrum’s Pearl app store, designed to optimize liquidity farming opportunities in DeFi, under its $1M Accelerator program.
• Sahara AI launched its incubator program at ETHDenver 2025, focusing on supporting AI and Web3 founders building decentralized AI solutions.
• Gensyn AI introduced Diverse Expert Ensembles, a model architecture designed for decentralized networks, emphasizing parallel training to improve performance through expert diversity.
• STP Network partnered with ElizaOS, to integrate Eliza’s AI-driven agents into the Autonomous World Engine, enabling complex multi-agent simulations for next-generation virtual worlds.
• Flock introduced Web3 Agent Model that leverages decentralized training and community collaboration, building on the AI Arena Task 1 to enhance blockchain-specific tasks like DeFi and smart contract analysis.
• Recall testnet is live, offering an AI agent competition platform that uses crypto incentives to drive improvement, benchmarking, discovery, and reward to the public.
• Inference Labs partnered with Terminus Network to enhance AI and blockchain integration with real-world payments.

That’s it for now.

– Marc & team