Keep the AI plans your team pays for. Govern every request.
Claude Code stays on your Claude plan and Codex on your ChatGPT plan, and coding plans from Z.AI, Kimi, MiniMax and more work as providers. The plan keeps paying for the tokens. Tempr adds what you'd have on an API key: request logs, limits, guardrails, and a model to carry on with when a plan runs out.
Plan requests count toward your Gateway plan's request quota, like any other request.
Whatever your team already pays for.
Claude plans
Claude Pro, Max, Team and Enterprise. Claude Code keeps its own sign-in: point it at Tempr and send your Tempr key as a header.
ChatGPT plans
Plus, Pro, Business, Enterprise and Edu. Codex keeps its ChatGPT sign-in through Tempr's Codex relay.
30 coding plans
Z.AI and Zhipu GLM, Kimi Code, MiniMax, Alibaba Cloud, Volcengine, Tencent, Xiaomi MiMo and more, each a provider of its own. Store the plan's key in the Portal, or send it with a request.
The plan pays. Tempr governs.
Every request, with the plan that paid for it.
Each request's page shows which plan or key paid and the plan's usage windows, and the logs filter by who paid. The dashboard's Coding plans card shows each plan's requests and what they would have cost at API prices.
- Which plan paid, on every request
- A "Paid by" filter on the logs, and in exports
- Each plan's value at API prices
Nobody's session stops when a plan is used up.
Give a key a model to use when a plan says it's used up, or a dollar budget runs out. The request moves there instead of failing, and the log says so.
- A used-up model per key
- Works for plans and for dollar budgets
- Fallback chains for providers that fail
anthropic/claude-opus-5.5 on dana@'s Claude Max planzai-coding/glm-5.3 on the team's Z.AI coding planPlan logins are a setting, not a surprise.
Personal keys accept a plan login by default. Organization keys accept them once an owner or admin turns on Plan logins, for the organization or for one key, since it moves billing onto each member's own plan.
- Off by default for organization keys
- On for the whole organization, or one key at a time
- Coding plan keys sent with a request are never stored
One setting per tool.
From the setup guide. Your Tempr key starts with tvk_.
// ~/.claude/settings.json: keeps your Claude plan { "env": { "ANTHROPIC_BASE_URL": "https://api.temprhq.io", "ANTHROPIC_CUSTOM_HEADERS": "x-tempr-key: tvk_..." } }
# ~/.codex/config.toml: keeps your ChatGPT plan model_provider = "tempr-chatgpt" [model_providers.tempr-chatgpt] name = "Tempr Gateway (ChatGPT plan)" base_url = "https://api.temprhq.io/backend-api/codex" requires_openai_auth = true wire_api = "responses" env_http_headers = { "x-tempr-key" = "TEMPR_API_KEY" }
# any OpenAI-compatible tool: the plan's key is stored in the Portal curl https://api.temprhq.io/v1/chat/completions \ -H "Authorization: Bearer tvk_..." \ -H "Content-Type: application/json" \ -d '{ "model": "zai-coding/glm-5.3", "messages": [{ "role": "user", "content": "Say hi" }] }'
About plans.
Does Tempr resell or pay for plan usage?
No. Your plan keeps paying its vendor exactly as before. Tempr relays the request, applies your rules and records it; each request counts toward your Gateway plan's request quota.
Which tools can use a coding plan?
Any tool that talks to an OpenAI-compatible or Anthropic-compatible API: Claude Code, Cursor, Cline, Zed, Kilo Code, OpenCode and your own apps. The docs list which tools each plan accepts.
Is a plan's key stored?
Only if you save it in the Portal, where provider keys are encrypted at rest. A key sent with a request is forwarded to that plan for that one request and never stored.
Are plans allowed to be used this way?
Each plan's terms are its vendor's, and Tempr can't change them. Some coding plans limit use to particular tools or forbid proxies; the setup guide lists what each vendor says, so check yours.
Point Claude Code or Codex at Tempr today.
Your plan keeps paying. You see every request, which plan paid, and what it was worth.