## Connect an AI Model 1. Open an AI Employee, then go to Settings → Model. 2. Select Add model and choose your AI Model service. 3. For Claude or OpenAI, paste your API key. Genosyn loads the available models and recommends a current choice. Select Connect AI Model to test and save it. For a custom endpoint, enter its URL and model ID, then select Add model. 4. To use ChatGPT subscription access, choose ChatGPT sign-in, continue, then select Sign in with ChatGPT on its card. Genosyn tests a reply after sign-in. Use Make active to switch between connected AI Models. ## Provider kinds Three provider kinds cover every setup. API-key and custom-endpoint models talk straight to the model API from Genosyn's in-process agent loop. OpenAI subscription models use OpenAI's official Codex app-server; there is still no generic provider CLI to install or maintain. ### Anthropic (Claude) - **Credentials**: Paste an API key. - **Connects to**: Claude models — Opus, Sonnet, Haiku. ### OpenAI (GPT) - **Credentials**: API key, or trusted single-tenant ChatGPT subscription access. - **Connects to**: Models available to the selected OpenAI access method. ### Custom - **OpenAI-compatible** - **Credentials**: Base URL + model id, plus an optional key. - **Connects to**: Ollama, vLLM, llama.cpp, LM Studio, or any gateway. The `Custom` kind is the path for any self-hosted or gatewayed LLM that speaks the OpenAI API — see [Open-source LLMs](/content/docs/open-source-models/index.html) for that flow. ## Credentials Everything a model needs is entered in the app. There is no persistent per-provider config directory to manage: - **Anthropic**: Paste an Anthropic Console API key. The runner picks the newest compatible Claude model available to your account, or you can choose another. - **OpenAI API key**: Paste an OpenAI Platform API key for direct, usage-based access through Genosyn's in-process model loop. - **OpenAI subscription**: On a trusted single-tenant Genosyn deployment, complete ChatGPT device sign-in or paste a Codex access token from an eligible Business or Enterprise workspace. Genosyn runs this model through the pinned `@openai/codex` app-server. The standard Docker default runs it beside bubblewrap-isolated coding and repository work; a host without Linux user namespaces falls back to subscription Runs with no coding tools. - **Custom**: Paste a base URL and model id, plus an optional API key if your endpoint requires one. Genosyn tests a real tool-use reply before saving the endpoint. The loop then points every request at that endpoint. ### Connect and test an AI Model During onboarding, or under an AI Employee's Settings → Model, choose AI Model service and paste your API key. Genosyn loads the models your account can access and recommends a recent compatible model. You can select another from Model, or open Choose a model ID manually to enter an exact ID. A choice you make is preserved when the list reloads. Select Connect AI Model. Genosyn sends a small request through the same API that employee work uses and checks that the model can call a harmless test tool. The request uses a small amount of your API allowance. Only a successful test saves and activates the AI Model. If the key, model access, billing, or network needs attention, an inline error explains what to check; correct it and try again. A failed key replacement keeps the previous working credential. The OpenAI and Anthropic model lists do not declare a general default. Genosyn recommends from their live catalogs rather than shipping a fixed model version, and tests the chosen model before connecting. ChatGPT sign-in uses the default advertised for that workspace. Existing AI Models keep their selected IDs; onboarding does not silently upgrade them. To change a connected ChatGPT AI Model, edit its model ID and save. Genosyn tests the new choice with the existing sign-in before saving. Leave the model ID blank to discover and test the workspace's current default. If the test fails, the previous model stays selected. ### Use an OpenAI subscription Subscription access is available on a trusted single-tenant deployment when coding execution is isolated with working Linux bubblewrap — the standard Docker default — or disabled. In the add-model form, choose OpenAI / ChatGPT, set Connect with to ChatGPT sign-in, then select Continue to ChatGPT sign-in. Genosyn uses your workspace's current default model and tests a real reply before marking it connected. ### Why Claude subscriptions are not offered Genosyn does not ask for or accept Claude.ai or Claude Code subscription credentials. Anthropic's [account authentication guidance](https://support.claude.com/en/articles/13189465-log-in-to-your-claude-account) tells developers building third-party products to use an API key and prohibits routing third-party traffic against subscription limits. Connect Anthropic with a Console API key instead. ## Context window Every turn sends the employee's Soul, their Skills, and their working set of tools, and each tool call adds its result on top. A long routine therefore grows until it reaches whatever the model will accept — so the direct agent loop needs to know how much room there is. When an API-key or custom-endpoint model connects, Genosyn asks the provider and shows the answer on the model card. For an OpenAI subscription model, the Codex app-server owns context management, so its card does not show the manual context-window controls. Once it knows, a run budgets against it: when the next prompt wouldn't fit, the oldest tool results are dropped to a stub so recent work and the routine's instruction survive. The run log says `[compact]` whenever that happens, so a forgetful-looking employee is always explained by its transcript. ## Built-in agent tools API-key and custom-endpoint models run through Genosyn's in-process agent loop; an OpenAI subscription model runs through the official Codex app-server. Both receive the same granted product and built-in browser catalogue. Subscription turns omit parallel delegation because they serialize on the model's credential-refresh lock. The installation mode controls coding tools and user-configured stdio MCP as described below. An employee is shown a small working set every turn and looks the rest up on demand; see [How tools reach the model](/content/docs/tool-discovery/index.html). ### How many tools an employee can hold The direct OpenAI API path accepts at most 128 tools on a request and rejects the whole turn if you send more. Anthropic publishes no such limit, and a custom endpoint sets its own. ## Multiple models & the active one An employee can hold several models side by side — say an `Anthropic` key for everyday work and an `OpenAI` subscription model for a second opinion. Exactly one is active at a time; the active model is the default brain for employee Chat and the model inherited by Routines and other AI surfaces. The most recently added model becomes active automatically — hit Make active on any other to switch, instantly and as often as you like.