reAPI is a unified HTTP API across image, video, chat, music and code models: one integration, OpenAI-style Bearer authentication, async task polling for generation jobs, and a credit system that reserves on submit and settles – with a refund – only at the job’s terminal state, so a failed generation does not quietly cost you. It advertises 99.96% uptime with automatic failover and states requests and responses are never stored. It is currently in beta.
Publishing the exact per-model rate – Seedance 2.5 shown at $0.071 a second, GLM-5.2 at $0.900 per million tokens, directly on the homepage – is more transparent than Requesty’s flat 5% markup, reviewed elsewhere in this catalogue: you can see the real unit cost of the specific model you intend to use rather than a percentage applied to an unlisted base rate. A stated zero-logging policy is also a meaningful commitment for anything generating content you would not want retained.
Beta status on infrastructure something else depends on is worth taking seriously – test failover behaviour and confirm the zero-logging claim against your own compliance needs before routing production traffic through it. As with any gateway, an outage here affects every model behind it at once, not just one provider.
