Unlimited Claude for your AI agent
Choose a timed access plan, connect your tool, and test a short request before starting a longer agent session. Keep the key group, API format and model aligned.
Available trials
Review the current offer and eligibility before continuing.
Choose access for the job
A timed plan covers eligible usage for its active period under fair-use limits. Prepaid credits suit occasional requests and work that does not fit a fixed session.
Current Claude Unlimited group rate limit: 60 requests per minute. Model availability and your selected plan's expiry still apply.
| Your workload | Start here | Before you start |
|---|---|---|
| A short test or one focused job | One-hour paid plan | Prepare your tool before the access period begins. |
| Regular coding or agent sessions | Compare timed plans | Check duration, eligible group and limits. |
| Occasional API calls | Prepaid credits | Check the model's current usage pricing. |
New paid access starts when it is granted to your account. Extending active access adds time to its existing expiry. Your account shows the exact end time.
Choose your tool
| Client configuration | Base URL | Key setting |
|---|---|---|
| Claude Code | https://aiprimetech.io | ANTHROPIC_AUTH_TOKEN |
| Anthropic Python or TypeScript SDK | https://aiprimetech.io | api_key or ANTHROPIC_API_KEY |
| OpenCode's provider/npm example | https://aiprimetech.io/v1 | Provider options.apiKey |
| Hermes custom Anthropic provider | https://aiprimetech.io/v1 | Provider key_env |
| OpenClaw custom Claude provider | https://aiprimetech.io | Provider apiKey |
Follow the complete guide for your installed version. Provider format and path handling differ between clients.
Using Codex CLI with GPT models? Follow the Codex setup guide; its Codex group is separate from Claude Unlimited.
Get one successful response
- Confirm your access is active. For a timed Claude plan, create or select a key in the Claude Unlimited group.
- Check the available models using the same key your agent will use.
- Follow the tool's setup guide and choose an exact returned model ID.
- Send a short prompt. Start the larger job after you receive a successful response.
curl -sS https://aiprimetech.io/v1/models \
-H "Authorization: Bearer $CLAUDEAPIKEY"Keep CLAUDEAPIKEY in your local environment. The following Messages request uses a small output limit; select a model returned for your key.
curl https://aiprimetech.io/v1/messages \
-H "x-api-key: $CLAUDEAPIKEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{"model":"claude-fable-5-1","max_tokens":64,"messages":[{"role":"user","content":"Reply with ready."}]}' Using Claude Fable 5.1
For timed Fable 5.1 work, select the Claude Unlimited group. Confirm claude-fable-5-1 in the model list for your key and send a short test request. The plan's rate limits and expiry still apply.
Fable 5.1 pricing and model details · Plan coverage and billing
If the request fails
| What happened | Check next |
|---|---|
| Key rejected | Check the key, active status and authentication format. |
| Model missing or unavailable | Check the key group and its available model IDs. |
| Access period ended | Check expiry in your account and choose another plan if needed. |
| 404 from a client | Check the base URL and /v1 suffix. |
| 429 during parallel work | Reduce parallel requests and follow rate-limit guidance. |
| 500 or 529 upstream error | Read the error and retry guidance in API troubleshooting. |
Keep the request ID, model ID and redacted error when contacting support. Do not share your key.
Making a tutorial or review?
Read the creator programme's participation and disclosure rules. Available offers and application status are shown only when a programme is open.
AI Prime Tech is an independent API gateway and is not affiliated with or endorsed by Anthropic. Timed access remains subject to the selected group, rate limits and expiry.
Codex and GPT API guides: Codex API key and access · OpenAI-compatible API · Codex CLI setup · GPT-6 model comparison