APIs and development
Any LLM API: Models, a Free Start, and AI Agents
A unified API lets you connect different models without configuring each provider separately. In Any LLM, one account manages API keys, the model catalog, and request history. Compatible clients connect through a familiar interface, and you select a model by its ID. Availability and terms change, so check the current catalog and limits before you start.
One endpoint and catalog instead of separate integrations
If an app or agent supports an OpenAI-compatible API, it usually needs only the service endpoint, a key, and the exact model ID. You can switch models in a request while keeping the same integration. Any LLM also supports Responses and Anthropic Messages for clients that need those formats.
The model list can change with availability and service settings. Get current IDs from `/v1/models`, and check prices and limits separately. A model appearing in the general catalog does not mean every route is free.
What is included in Any LLM’s free start?
After signup, a free account gets up to RUB 10 in combined request credit for GPT-6 Luna, Claude Haiku 5.5, GLM 5.3 Flash, and DeepSeek V4.1 Flash 1017. This is one shared starter allowance across those models, not RUB 10 for each model.
The catalog may also include zero-priced routes. Any LLM applies a shared limit of 30 requests per user per UTC day for free models. Provider limits also apply, so check each model’s current price and availability in the catalog.
One key for popular model families
You do not need a separate key from every model developer or a new client setup each time you switch. Get an Any LLM key, request `/v1/models`, and select an available model: GPT for everyday work and code, Claude for long materials, Gemini for multimodal tasks, and Qwen or Kimi for coding and specialized work.
The available catalog appears in the chat and API before you send a request. Specify the exact model ID; its request history is saved in your Any LLM account.
Compatible API formats
The base URL is `https://api.any-llm.online/v1`. Schemas and current routes are in the interactive API documentation. One account manages models, API keys, request history, and spending.
| Format | Use case |
|---|---|
| Chat Completions | Chat apps and most OpenAI-compatible SDKs |
| Responses | Apps using the Responses API and continuation with a saved ID |
| Anthropic Messages | Clients that connect to the Messages API |
Connect Hermes and other agents
Create a key in the API dashboard, then configure Hermes, OpenCode, Codex CLI, or Claude Code. To set up Hermes manually, choose Custom endpoint, enter `https://api.any-llm.online/v1`, and add your key. The key is shown when created; store it like a password and revoke it if it appears in public code or logs.
Open the API page, then go to API key settings. You can review current models, create a separate key, and check spending there.
Ruble billing, shared budgets, and spending history
MAX costs RUB 1,990 per calendar month and includes RUB 1,600 in monthly AI request credit. The same budget covers both chat and API, so there is no need for a separate API subscription. PRO has a shared weekly budget. If your subscription allowance is insufficient or you choose balance billing, your account balance applies.
Before a request, the system estimates and reserves funds for the full context and output limit. It releases the unused reserve after the response; the cost and status appear in API history. You can top up and subscribe in rubles through the account dashboard.
Make your first request
Create a key in API settings, copy it once, and save it in a secrets manager. Request `/v1/models`, choose an available ID, and send a test request to `/v1/chat/completions`. For repeatable operations, set a unique `Idempotency-Key`. Do not automatically start a new request if the previous result is unknown.
The developer guide includes curl and client examples, history, budgets, and configuration details.