Use your Claude subscription as an API

Ways to call Claude from your own code

A Claude plan on its own gives you no API key. The table below compares the four ways to call Claude from your own code.

Anthropic's help center states: "The Pro plan does not include API usage through the Claude Console." Max 5x and Max 20x include $100 and $200 a month in Claude API credits, claimable after 7 days on the plan by linking a Console organization under Settings > Billing, and unused credits expire at the end of every billing cycle. Plan2API is the option that runs on your plan's own usage limits, with no separate Console balance (see the API vs. subscription cost breakdown for the math).

Connect Claude Pro or Max in 3 steps

1. Sign up at plan2api.com/connect with a Google account.
2. Connect your subscription by signing in with your Claude account on Anthropic's own sign-in page.
3. Copy the p2a_ API key generated for you in the dashboard, where the Quickstart page has copy-paste setup for curl, Python, and Node.

The curl example below streams a response from claude/claude-sonnet-5 through https://api.plan2api.com/v1/chat/completions, sending the p2a_ key in an Authorization: Bearer header because the x-api-key header is rejected.

curl https://api.plan2api.com/v1/chat/completions \
  -H "Authorization: Bearer $PLAN2API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude/claude-sonnet-5",
    "stream": true,
    "messages": [{"role": "user", "content": "Explain rate limits in one sentence."}]
  }'

Claude models you can call

The table below lists all 9 Claude model IDs available on Plan2API with their context windows and output limits, alongside what the same calls cost per million input and output tokens on the pay-per-token Claude API.

• The anthropic/ prefix and dotted model names such as claude-opus-5.5 work too.
• GET /v1/models returns the valid model ids.
• Anthropic's newest Sonnet 5.5 and Haiku 5.5 models are not in the list.

What works and what doesn't

• Streaming and non-streaming responses both work on POST /v1/chat/completions.
• Tool and function calling work.
• Reasoning effort works through reasoning_effort or OpenRouter-style reasoning.effort.
• Image input works, and image URLs in your messages are fetched for you (up to 32 images per request, up to 20 MB and 15 seconds each).
• Both max_tokens and max_completion_tokens are accepted, with request bodies up to 64 MB.
• The API returns Access-Control-Allow-Origin: * CORS headers so browser apps can call it directly.
• temperature and top_p parameters are ignored on Claude models.
• There is no Anthropic /v1/messages endpoint, and no /v1/responses, /v1/completions or /v1/embeddings.

The Python example below calls claude/claude-opus-5-5 with high reasoning effort through the official OpenAI SDK.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.plan2api.com/v1",
    api_key="p2a_...",
)

reply = client.chat.completions.create(
    model="claude/claude-opus-5-5",
    reasoning_effort="high",
    messages=[{"role": "user", "content": "Review this function for bugs: ..."}],
)
print(reply.choices[0].message.content)

Usage limits per Claude plan

Plan2API adds no token quota of its own, so your Claude plan's rolling 5-hour session and weekly limits apply alongside your Plan2API plan's request limits. The table below breaks down how usage works across Claude Pro and Max.

Anthropic publishes no fixed message counts for its plans, and all activity across web, desktop, mobile, and Claude Code draws from the same pool. You can check Settings > Usage on claude.ai to see how much of your 5-hour session and weekly limits you have used, or turn on usage credits on paid Claude plans to keep working past a limit at standard API rates.

If you connect two or more Claude accounts to Plan2API, a request that hits a limit on one account moves to the next connected account before any output is sent. When every connected Claude account is out of quota, the API returns HTTP 429 subscription_quota_exhausted with a Retry-After header in seconds for the earliest reset.

Use your Claude plan in scripts and apps

One p2a_ key covers your Claude plan and can also route to ChatGPT (Codex), Google AI Pro and Ultra (Gemini through Antigravity), and Kimi (through Kimi Code) subscriptions, selected by model prefix (claude/, openai/, gemini/, or kimi/). You keep paying Anthropic for your Claude plan, and Plan2API never bills per token.

Connected logins are encrypted with HPKE (X25519) and AES-256-GCM so only your API key can unlock them, and the key's secret is never stored. Plan2API never stores your prompts, outputs, files, or images, runs its servers and database on OVHcloud in Germany, and has you sign in on Anthropic's own page so it never sees your Claude password (how Plan2API protects logins).

• Free is $0 with no card needed and includes 1 connected subscription, 2 API keys, and 50 chat requests a day.
• Pro is $12 a month or $99 a year and includes 3 connected subscriptions, 10 API keys, no daily request cap, and image generation.
• Scale is $29 a month or $249 a year and includes 25 connected subscriptions, 50 API keys, no daily request cap, and image generation.