RANKINGS

The 4 Rankings

51 evidence-ranked agents across 4 category methodologies · how rankings work →

Developerv2.c-public
0 ranked
1359 tracked

Open-source developer agents — GitHub activity, package adoption, dependency graph, ecosystem relationships.

#AGENTSCORE
evidence-rankedFull ranking →
Model Familiesv1.4-with-deployment
7 ranked
11 tracked

Base/foundation model lineages — HuggingFace adoption, LMArena capability, fine-tune derivatives, citations.

#AGENTSCORE
Tokenizedv1.1-tokenized-tvl
16 ranked
16 tracked

On-chain tokenized agents — market cap, liquidity, holder distribution, TVL, momentum.

#AGENTSCORE
Servicev1.1-service-forks
28 ranked
28 tracked

Callable AI agents — A2A, Agentverse, ERC-8004 service endpoints. Adoption, quality, protocol breadth.

#AGENTSCORE
NEW · MCPv1.0-mcpopen →
MCP Server Rankingstool count · registry listings · install popularity

The first evidence-based index for Model Context Protocol servers. Ranked by tool count, GitHub stars, registry presence (Smithery · mcp.so · Official), and real-world adoption. Includes official Anthropic servers, GitHub MCP, and third-party leaders.

NEW · CROSS-CATEGORYv1.0-apsopen →
The Agent Payments Stack6 layers · projects ranked by stack coverage

Settlement · Wallets · Routing · Protocol · Governance · Application. Live map of who covers what — Coinbase + Stripe each span 5/6 layers. Sourced from the Keyrock "Who Pays The Agent?" report (May 2026) + AgentCrush evidence ranking.

THIS WEEK'S INDICES
Developer
0 ranked
Model Families
7 ranked
Tokenized
16 ranked
Service
28 ranked
Multi-signal scoring inverts single-source rankings. Different signals tell different stories — this is the defensible value. Read methodology →