Pricing
List prices ($ per million tokens) scraped from each provider's public catalog, plus a live "What's 2+2?" probe that confirms the published rate matches the billed usage.
| Provider | Model | $/M in | $/M out | prompt tok | compl tok | this call $ | lat | err |
|---|---|---|---|---|---|---|---|---|
no pricing snapshot — run python -m probes.collect --pricing to populate
| ||||||||
Subscription math
Several providers offer flat-rate subscriptions that don't map cleanly onto $/M:
- RedPill Pro ($50/mo). Used to be a credit pool; 2026-04-21 measurement gave a
sub/list ratio = 0.998— effectively no discount vs the pay-as-you-go rates above. - Venice Pro ($18/mo). Chat-UI only for API — provides $1 of API credit. Pro+ ($68/mo) gives $75 credit (~10% bonus). Max ($200/mo) gives $225 (~12.5% bonus).
- Z.AI GLM Coding Plan. Separate coding endpoint (not TEE-attested). Pro $30/mo works out to roughly $0.08–0.15 $/M on GLM-5.1 — about 30× cheaper than RedPill's list GLM-5.1 but with no attestation.
These rates are observations from manual runs; they are intentionally not scraped into the daily table above because they require subscription authentication and the math is sensitive to the reference workload.