SemiAnalysis built a subscription‑dashboard that quantifies the real‑world value of AI lab plans by measuring how each token type (input, cache write, cache read, output) moves a provider’s usage meter. Experiments isolate one token type per run, record the meter’s incremental steps, and derive a per‑token rate after discarding partial steps and subtracting fixed‑overhead tokens. Rates are converted to tokens per plan window (5‑hour or 7‑day limits) and then to dollar equivalents using the plan’s monthly price and an agentic workload shape drawn from the Tokenomics Model. Applying this method shows that a $200/month Claude plan delivers roughly five times the API‑equivalent value of OpenAI’s comparable $200 plan when running mid‑tier models such as Opus 5.5 versus GPT 6.1 Sol, because Anthropic’s credit‑cost ratios remain far more favorable. OpenAI’s recent halving of the $200 plan’s API‑equivalent value (and the introduction of a $500 tier that adds only ~21% more Astra tokens) narrows the gap but still leaves Anthropic ahead. The dashboard also tracks Meta, SpaceXAI, Cursor, Cognition, Z.ai, MiniMax, and Moonshot, updating limits daily to capture silent A/B tests or promo‑driven changes, thereby providing a granular, workload‑aware view of subscription economics across the AI ecosystem.