Questions about unmetered access and billing.

Yes. OpenLimits supports OpenClaw, Hermes, Pi, Conductor, OpenCode, Cursor, Claude Code, Codex Desktop, compatible SDKs, and direct API clients.

Every plan includes Opus 5.5 and Opus 5, Sonnet 5, Haiku 4.5; GPT-6.1 Sol, GPT-6 Sol, Astra, and Luna; GLM 5.3, 5.2, and 5 Turbo; MiniMax M3; and DeepSeek V4 Pro and Flash. Plans differ by renewal period, fair-use ceiling, priority, and support—not model availability.

Yes. Set the compatible base URL and auth token, then use Claude Code normally.

Yes. Every plan includes the GPT and Codex-compatible models. The same key and dashboard apply.

You need an API key and a compatible base URL. Most clients can be configured in a few minutes.

Yes. There is no per-token charge and no credit balance to top up. Day, Week, and Max apply a fair-use ceiling of $2,000 per day in API-equivalent usage; Ultimate has no daily ceiling. The ceiling is there to prevent resale, and normal heavy development does not approach it.

There are no hidden fees and no usage-based charges. The plan price is the entire bill. Plans renew until you cancel, and time already used is not refundable.

At provider list prices the Max fair-use ceiling is worth roughly $60,000 of API usage a month, against €120 for the plan. Your dashboard tracks the API-equivalent cost you have avoided in real time.

Further requests are declined until the daily window resets at 00:00 UTC. It is a ceiling, not an overage charge — you are never billed more than your plan price.

Day, Week, Max, and Ultimate renew on their listed schedules until you stop renewal from the billing portal.

Future renewals stop. Access remains active until the end of the paid period, and dashboard history stays available.

Two reasons. Provider capacity is pooled across users, so one customer does not fund an entire account at peak capacity, and you pay a flat subscription instead of per-token rates that rise with every hour you work.

A global guard admits 30 request starts per second — far above what any interactive coding agent generates. There is no pool concurrency cap for work already in flight, and no cap on how many tokens those requests consume.

OpenLimits does not store prompt or completion content. It records routing and usage metadata required for operation, billing, and dashboard analytics.

Control-plane work is designed to complete before streaming begins. End-to-end latency still depends on the selected upstream provider.

Yes, subject to the plan terms and normal upstream availability. The public APIs preserve compatible request and streaming behavior.

Yes. Send supported media and tool schemas through the corresponding provider-compatible API format.

OpenLimits can retry eligible transient account failures. A provider-wide outage still prevents models on that provider from completing requests.