Plans for every agent.
OpenLimits gives OpenClaw, Hermes, Pi, and your own agents one API key for GLM, MiniMax, DeepSeek, Claude, and Codex model groups.
| Model access | CORE$39/ moOpen model access for lightweight agents and background jobs | AGENTS$79/ moThe default plan for day-to-day coding agents | WEEK PASS$45/ weekFull Max access, billed weekly — every model group, no limits | DAY PASS$10/ 24hFull Max access, billed daily — every model group, no limits | MAX$120/ moFull access for heavy agent workloads and premium models |
|---|---|---|---|---|---|
| Model groups | |||||
| Unlimited GLM 5.1 & GLM V5 Turbo | ✓ | ✓ | ✓ | ✓ | ✓ |
| Unlimited MiniMax M3 | ✓ | ✓ | ✓ | ✓ | ✓ |
| Unlimited DeepSeek V4 Family | ✓ | ✓ | ✓ | ✓ | ✓ |
| Unlimited Claude Sonnet + Haiku | — | ✓ | ✓ | ✓ | ✓ |
| Unlimited Codex mini | — | ✓ | ✓ | ✓ | ✓ |
| Unlimited Claude Opus | — | — | ✓ | ✓ | ✓ |
| Unlimited full Codex/GPT group | — | — | ✓ | ✓ | ✓ |
| Limits and routing | |||||
| No cooldowns | — | No caps | ✓ | No caps | ✓ |
| No 5-hour rolling limits | — | — | ✓ | — | ✓ |
| No 7-day usage caps | — | — | ✓ | — | ✓ |
| Automatic failover | ✓ | ✓ | ✓ | ✓ | ✓ |
| Highest model-group access | — | — | — | — | ✓ |
| API features | |||||
| Streaming responses | ✓ | ✓ | ✓ | ✓ | ✓ |
| OpenAI-compatible endpoint | ✓ | ✓ | ✓ | ✓ | ✓ |
| Works with Claude-compatible clients | ✓ | ✓ | ✓ | ✓ | ✓ |
| Effort levels + extended thinking | — | ✓ | ✓ | ✓ | ✓ |
| Image support | ✓ | ✓ | ✓ | ✓ | ✓ |
| Dashboard & analytics | ✓ | ✓ | ✓ | ✓ | ✓ |
| Usage history | ✓ | ✓ | ✓ | ✓ | ✓ |
| Billing | |||||
| Cancel anytime | ✓ | ✓ | ✓ | ✓ | ✓ |
| Start with | Start Core → | Start Agents → | Start Week Pass → | Start Day Pass → | Start Max → |
Model groups without client rewrites.
Start agents on efficient open models, move harder work to Claude or Codex, and keep the same endpoint. See all features.
Open + premium model groups
Use GLM, MiniMax, and DeepSeek for fast agent loops, then route premium work to Claude and Codex from the same OpenLimits key.
Zero Rate Limits
No 5-hour rolling limits, no 7-day caps, no cooldowns. Code all day without interruption. Automatic failover ensures your requests always go through.
Full Dashboard
Real-time analytics, token tracking, model breakdowns, cost estimates, and a live request feed. Know exactly how you're using your API access.
Native agent support
OpenClaw, Hermes, Pi, Conductor, Cursor, OpenCode, and direct SDK clients can all point at the same endpoint.
All Advanced Features
Extended thinking, effort levels, streaming, tool/function calling, PDFs, images — every Anthropic API feature works out of the box.
Works Everywhere
Claude Code CLI, Codex Desktop, Conductor, Cursor, OpenCode, OpenClaw, Hermes, Pi, Python/TypeScript SDKs, cURL, and compatible clients. One key for everything.
Common pricing questions
Is there a free tier?
No, but the $10 Day Pass gives you a full day of Max access to evaluate the service — every model group, every feature, no limits. Cancel anytime before it renews.
Can I cancel anytime?
Yes. Manage or cancel your subscription anytime from the billing portal in your dashboard. When you cancel, your access continues until the end of the period you've already paid for, then stops renewing. Your dashboard data carries over.
Is it a subscription?
Yes. All plans are recurring subscriptions that renew automatically until you cancel. Day renews every 24 hours, Week renews weekly, and monthly model-group plans renew monthly. Cancel anytime in the billing portal.
Why is it so cheap compared to the API?
We pool resources efficiently across users. You get API-level access at a fraction of the per-token cost. The math works because not everyone is coding at peak capacity 24/7.
Can I get a refund?
Full refund within 14 days if you haven't used the service at all. After any API usage, all sales are final. See our Refund Policy for details.
Do you offer team plans?
Not yet. Each account gets one API key. For teams, each member needs their own account. Team plans with shared billing and admin controls are on the roadmap.