> ## Documentation Index > Fetch the complete documentation index at: https://docs.karta.sh/llms.txt > Use this file to discover all available pages before exploring further. # BYOK - bring your own key > Use your own model-provider key or compatible endpoint, select a model and portable parameters, and rotate the connector when needed. By default, turns run on Karta's platform model key. **BYOK** lets an org supply its *own* provider key so usage bills to your provider account and runs under your own rate limits and contracts. BYOK is a paid-plan capability; which plans include it is shown in the console. Managed plans run on the platform key and prepaid [credits](/platform/billing-and-credits). ## Providers You can add multiple connectors per organization for `anthropic`, `openai`, `bedrock`, `vertex`, `openrouter`, `xai`, or `google`. Provider identity and API format are separate: an xAI or Google connector remains labeled as that provider even when it uses an OpenAI-compatible API. Choose a provider connector that your [supported harness](/build/supported-harnesses) can use. Readiness-gated versions of Claude Code, OpenCode, DeepAgents, Goose, and Codex CLI share the same maintained model-routing contract. Each can use the current Anthropic, OpenAI, Google, and xAI frontier routes, plus a custom OpenAI-compatible endpoint and model ID. The selected harness version publishes its model-readiness status and capabilities in the harness catalog. ## Custom endpoints Set an HTTPS endpoint and explicit API format when your provider exposes an OpenAI-compatible API or you operate a compatible gateway: ```bash theme={null} echo "$XAI_API_KEY" | karta model-keys create xai production-xai \ --key-stdin \ --endpoint-url https://api.x.ai/v1 \ --api-format openai_responses echo "$COMPAT_API_KEY" | karta model-keys create openai private-models \ --key-stdin \ --endpoint-url https://models.example.com/v1 \ --api-format openai_chat_completions ``` The endpoint must use public HTTPS and cannot contain credentials, a query string, or a fragment. Karta validates the destination again when connecting. Your prompts, tool definitions and results, and model output are sent to the selected endpoint under that provider's terms. ## Choosing models Each key offers a set of models. Pick an org **default** - a key plus a model from it - on the dashboard. Every agent uses that default unless it overrides to a different key and model, so one org can route different agents to different keys. Custom model IDs are passed through as provider-owned opaque strings; they do not need to appear in Karta's managed model catalog. ## Model parameters Organization defaults and agent overrides support the same portable controls: | Parameter | Accepted value | | ------------------- | ----------------------------------------------------------------------------------------- | | `temperature` | Finite number from `0` to `2`. | | `top_p` | Number greater than `0` and at most `1`. | | `max_output_tokens` | Positive integer, capped at `1000000`; provider/model limits may be lower. | | `reasoning_effort` | `low`, `medium`, or `high` on supported OpenAI-compatible OpenAI, xAI, and Google routes. | An agent can override one field while inheriting the rest from the organization. Unsupported route or parameter combinations fail before the harness launches; Karta does not silently drop parameters or switch providers. ## Anthropic subscription Instead of an API key, you can connect an Anthropic subscription. Use it for your **own** coding work running on Karta - not to serve requests on behalf of your end-users, which Anthropic's terms do not permit and which can lead to your subscription being suspended. For any agent serving users, virtual employees, backend jobs, or customers, use an API key. ## Store a key Set one from the dashboard, CLI, or management SDK: ```bash theme={null} echo "$ANTHROPIC_API_KEY" | karta model-keys create anthropic primary --key-stdin karta model-keys list karta agent model support-bot \ --connector-id \ --model claude-sonnet-5 \ --temperature 0.2 \ --max-output-tokens 2048 karta agent model support-bot --source inherit karta model-keys revoke ``` ```typescript theme={null} import { Karta } from "@karta.sh/sdk"; const karta = new Karta({ apiKey: process.env.KARTA_API_KEY! }); await karta.modelKeys.create({ provider: "xai", name: "production-xai", plaintext: process.env.XAI_API_KEY!, endpoint_url: "https://api.x.ai/v1", api_format: "openai_responses", }); await karta.modelKeys.list(); await karta.modelKeys.revoke(id); ``` ## How it is used at request time Karta resolves a connector and model once for each turn: | Selection | Result | | ------------------------------------ | -------------------------------------------------------------------------------------------------------- | | Agent pins a connector | The agent uses that connector and its selected model. | | Agent inherits from the organization | The organization default connector and model are used. | | No connector is selected | The platform connector is used. | | A selected connector is revoked | Turns fail until the selection is changed; Karta does not silently route them through another connector. | If a provider rejects a BYOK key, the turn surfaces a `byok_key_rejected` error * `422` for unary requests, or an [`error` event](/concepts/streaming-events#error) mid-stream for streaming requests. Invalid or unsupported model settings return stable `model_route_invalid`, `model_route_unsupported`, or `model_parameter_unsupported` errors before a provider request is made. ## Safety BYOK plaintext is **never logged, never put on spans, and scrubbed from error reports**. Karta processes customer content to run your agent, under the operator-access policy and audit posture described in [Trust](/security/trust).