Crypto Al Benchmark Alliance
Defining Crypto-Native Al Benchmarks
CAIBA’s mission
CAIBA unites leading crypto innovators to create open, unbiased scoreboards and open-source toolkits that measure how agents think, plan, and execute on-chain.
Through ongoing knowledge, planning, and action benchmarks, we provide all builders with clear standards and tools for improvement, helping the entire ecosystem advance together.
Founding Members
Cyber
Cyber combines AI and social context to make crypto easier to use, understand, and build on.
Alchemy
Your complete developer platform to build AI agents that manage wallets, execute secure transactions, and move assets across every chain.
Eigenlayer
On-chain compute infra for AI workloads
RootData
Comprehensive blockchain data for model training
Goldsky
High-performance indexing via subgraphs and realtime data streaming pipelines.
Sentient
AI research network for decentralized intelligence.
IOSG Ventures
Industry leading Research ventures
Newton
Crypto UX is broken. AI can't be trusted. Newton sets you free. Simpler crypto UX with verifiable...
Thirdweb
Dev tools for seamless AI & blockchain intelligence
Metis
AI Aligned, Human Defined.
OpenGradient
The L1 for AI Models
MyShell
The first AI consumer layer - build, share, and own AI Apps.
Surf
Crypto insights at your fingertips, a shortcut away, no tabs, no fluff, just alpha.
LazAI
LazAI is a Web3-native AI network redefining data for the AI era - making it verifiable...
Kite AI
Architecting a purpose-built L1 for the agentic internet with native support for real time payments, programmable governance, and cryptographic trust.
DMind AI
An open-source AGI institute bridging real world crypto data with foundational AI research and open benchmarks.
Buzzing
Real-time betting engine and truth oracle across social channels, turning noise into insight.
FLock
Pioneering decentralized AI training via federated learning on blockchain rails.
Nexus
A world supercomputer to enable the AI economy.
Ormi
A hyperscale data platform delivering live, historical, and AI-enriched data at sub-second latency.
Codatta
Building a decentralized Knowledge Layer, tailored for post-training agent fine-tuning through high-quality community data.
GM Agents
All your AI agents in one app — use more, earn more.
CAIA Benchmark
(Crypto AI Agent Benchmark)
CAIA gauges whether a model can shoulder the day-to-day work of a junior crypto analyst. It presents agents with real-world tasks - onchain transaction analysis, project discovery, and tokenomics diagnostics—demanding not only factual knowledge but coherent planning, blockchain-native reasoning, and faultless tool use. A high CAIA score indicates that an agent is ready to contribute to professional crypto research while giving builders a transparent yardstick for further improvement.
Wish to join CAIBA?
or Direct message @James_Dai on telegram
Access secure, unbiased benchmarks to validate your crypto AI agents against real-world data.
✅ Transparent, reproducible scores
✅ Live onchain evaluation
✅ Community-reviewed protocols