Momentum
Claude leads PostHog's MCP traffic, but bad arguments dominate failures
Community reports 2 sources
PostHog's public view of model and harness usage, tool calls and errors on its own MCP server, with explicit identification and sampling limitations.
What happened lately
PostHog publishes a leaderboard from tool-call events on its own Model Context Protocol server. It separates model usage, client harnesses, tool categories and error rates. The October 15 snapshot covers October 5-11 for its weekly comparisons.
Most model labels are supplied by agents themselves, and some calls have no model label. These observations describe PostHog’s traffic rather than global model adoption or a controlled benchmark. Server-side latency also excludes model thinking and network time.