Compare

One model, three supplies in one coordinate system: official API, relay API, subscription account or top-up. The first two are priced per million tokens, the third per month — comparing them needs a monthly-usage assumption, and that assumption belongs next to the number, not hidden.

The subscription column has no data yet. The collector is not connected, so only the first two supplies are shown. Drawing an empty frame and pretending is worse than missing a column.

ModelOfficial API $/MRelay low $/MRelay median $/MLow / officialSubscription
GPT 5.6 Sol5.000.1501.000.03×not connected
GPT 5.6 Terra2.000.0600.4000.03×not connected
GPT 5.55.000.1501.000.03×not connected
GPT 5.42.500.0750.5000.03×not connected
Grok 4.52.000.0600.1200.03×not connected
Opus 55.000.3001.750.06×not connected
Opus 4.85.000.3001.750.06×not connected
Sonnet 52.000.1200.7000.06×not connected
Haiku 4.51.000.0601.000.06×not connected
GPT 5.6 Luna0.2000.0150.0400.07×not connected
Gemini 3.5 Flash1.500.1380.2250.09×not connected
Opus 4.75.000.5005.000.10×not connected
Opus 4.65.000.5005.000.10×not connected
Sonnet 4.63.000.3003.000.10×not connected
Fable 510.001.503.500.15×not connected
Kimi K33.001.502.860.50×not connected

Official prices are each vendor's published input price, in USD, with no conversion. Relay prices come from model_ratio × group_ratio × $2/M (new-api's 1 ratio = $0.002/1K); groups on expression-based billing are excluded rather than guessed.