Accurate local-LLM memory math — will this model fit your GPU or Mac? Models sliding-window, hybrid/linear, MLA compressed-KV and MoE architectures
Accurate local-LLM memory math — will this model fit your GPU or Mac? Models sliding-window, hybrid/linear, MLA compressed-KV and MoE architectures that naive VRAM
This is the open calculation core of FitLLM. The math is open so you can audit it.
Compare MCP servers by category, or browse hosted MCP servers if you prefer an endpoint over a local process.
{
"mcpServers": {
"fitllm": {
"command": "npx",
"args": ["-y", "fitllm"]
}
}
}Add to claude_desktop_config.json, then restart Claude Desktop.
Store entities, relations and observations in a local knowledge graph with the reference Memory MCP server.
Connect the Notion MCP server to search workspace content and work with pages and databases through Notion’s hosted OAuth service.
Find business leads and live opportunities for a business, with the proof and the next step.
Social listening for AI agents: search mentions, pull scored buying signals, score leads
Hosted CRM and task desk for AI agents: per-agent keys, scopes, pause and audit log.
LeadConnector / GoHighLevel MCP Pack — wraps the GoHighLevel CRM for AI agents.