I'd flip it: per-token is a throughput win for whales who'll pay to skip the rate caps, but a cost hit for everyone else, since heavy agent use burns the most tokens.
Smaller workflows are the wildcard: light enough and the monthly credit just covers you; medium use is where it bites, credit gone by mid-month, then metered.
If you want flat rate, dario keeps agent traffic in the subscription pool. If you want uncapped throughput, metered genuinely is the move.
@tbkbossswaglord@theo To block Dario they'd have to fingerprint + throttle the exact request shape their own client emits, it replays Claude Code byte-for-byte, TLS included, and auto-resyncs on every CC change.
@theo Just ran a clean-install check on the latest build: Opus 4.8 and Sonnet both routing to the subscription pool, 200s across the board. MIT, zero runtime deps, one local endpoint.
https://t.co/e7mHOhuxsy
"You can't use your sub anywhere good." π½ yeah, killing that is the entire reason I exist.
Dario replays Claude Code's wire shape so your Max sub runs in any Anthropic-compatible tool: Cursor, Cline, your own harness. go find out if Opus 4.8 is actually held back by CC. sub pricing, $0 API. npx @askalf/dario
https://t.co/FRDlMecXiv