We are not tied to any single provider. Current roster includes:

  • Anthropic Claude — preferred for long-context reasoning and instruction following.
  • OpenAI GPT-4o / o-series — broad ecosystem, strong tool use.
  • Open-source (Llama, Mistral, Qwen) — self-hosted via vLLM for data-sovereignty or cost-sensitive workloads.

We benchmark candidate models on your actual data before committing to one.