Getting Started

Choose the AI providers and models that power your Sordit Agents

Sordit supports OpenAI, Anthropic and OpenRouter. Activate one or all three, choose the models that suit your work, and pay for AI token usage directly through your own provider accounts.

Choose your AI providers

You can activate any combination of these providers in your Sordit workspace:

  • OpenAI: Supports the GPT-6 family, including GPT-6 Luna, Sol and Astra.
  • Anthropic: Supports Claude Sonnet 5.5, Opus 5.5 and Fable 5.1.
  • OpenRouter: Gives you access to a much wider selection of models from different AI companies.

Once activated, you can select the provider and model you want to use when creating each Sordit Agent.

OpenAI
Anthropic
OpenRouter

Bring your own AI provider account

Sordit doesn’t charge you for AI tokens or add a markup to your provider’s token pricing.

When you activate a provider using your own API key:

  • Your AI token usage is billed directly to your OpenAI, Anthropic or OpenRouter account.
  • You manage your model selection, usage limits and provider costs.
  • Sordit credits are charged separately for Sordit platform usage, including cloud Agent execution and applicable business-tool operations.

This keeps your AI model costs separate from the cost of running automations on Sordit.

Choosing models through OpenRouter

OpenRouter offers access to hundreds of AI models through a single API.

However, not every model is suitable for running Sordit Agents.

Sordit is designed for persistent business operations, where Agents may need to work through multiple steps, use business tools, make decisions and complete tasks reliably.

To help you choose compatible OpenRouter models, Sordit validates several requirements:

  • Text input: The model must support text-based instructions.
  • Reasoning: It must support reasoning settings compatible with Sordit’s Agent runtime.
  • Tool calling: It must be able to interact with external tools.
  • Structured outputs: It must support the required structured response formats.
  • Context window: At least 256K tokens.
  • Availability: A paid, non-free model route.

These checks help identify technically compatible models, but don’t guarantee reliable execution.

For more demanding workflows, we recommend evaluating capable models such as Kimi K3 or DeepSeek V4 Pro, rather than choosing a lightweight Flash model purely for its lower token cost.

A stronger model may be more economical overall if it completes the work with fewer retries. We recommend testing your chosen model with representative Sordit Jobs before using it for ongoing business operations.

Get started with Sordit →
← Back to In Action