Tracks Anthropic’s share of eligible token usage among models appearing in OpenRouter’s daily top 50, measured over the seven most recent complete, consecutive UTC days.
Index = 100 × Anthropic tokens ÷ eligible top-50 tokens
Prompt and completion tokens are summed across all seven days before calculating the percentage.
Anthropic tokens are identified using OpenRouter’s exact author attribution. OpenRouter’s aggregate “other” row, multi-model routers, and floating aliases are excluded. This measures public traffic routed through OpenRouter only—not Anthropic’s total API usage, cloud-provider usage, revenue, requests, users, or overall AI market share. Token counts use providers’ respective tokenizers, so cross-provider comparisons are approximate. Manifold refreshes the oracle hourly and pauses trading if the source becomes stale, incomplete, or invalid.
Source: https://openrouter.ai/rankings
Rankings data is licensed under CC BY 4.0.
1,000@Jack1 We use our own methodology to build the index/oracle price (explained in the description), but you can see their version of this here: LLM Rankings | OpenRouter, this uses more than just the top 50 and doesn't have any smoothing like ours does.
We do not have license to use their final calculation but we can use their raw figures to create our own index/oracle