Providers
Five providers. One agent loop.
Linda is provider-agnostic. The transport layer abstracts the wire format; your config picks the model. Swap with one line.
Anthropic Claude
Sonnet 4.6, Opus 4.7, Haiku 4.5 The default. Best agentic behavior, native tool calling, streaming, prompt caching.
Configure Anthropic Claude →OpenAI
GPT-4o, GPT-5, o3-mini Strong tool calling. Use when you have OpenAI credits and want to stay there.
Configure OpenAI →Groq
Llama 3.3 70B, Qwen 2.5 Fastest tokens per second on the planet. Best when latency dominates.
Configure Groq →OpenRouter
~50 models Aggregator. Useful for A/B testing models without changing code.
Configure OpenRouter →Ollama
Local — Llama, Qwen, Mistral, more Runs on your laptop. Privacy-first; no provider sees the traffic.
Configure Ollama →Ship an agent-driven flow this afternoon.
Install Linda, paste a config, and your form turns into an agent that fills its own inputs.