Adapters Overview
VernLLM's LLMClient interface is modeled on OpenAI's chat.completions.create, adapters translate other providers into that shape
VernLLM expects a client shaped roughly like OpenAI's chat.completions.create, its own
LLMClient interface, not a copy of OpenAI's actual SDK types. Some providers already match
closely enough to need no adapter; others need one that translates their native API into it. A
few fields (ContentBlock[] multimodal input, is_error on tool messages) are VernLLM's own
additions with no OpenAI equivalent, adapters translate those as needed per provider too.
OpenAI-Compatible
Groq, Mistral, DeepSeek, Cerebras, Together, Fireworks, Ollama, OpenRouter, Perplexity, DeepInfra, Novita, Hyperbolic, Moonshot, Zhipu, LM Studio, vLLM, and any other OpenAI-wire-compatible provider
Anthropic
fromAnthropic
Gemini
fromGemini
AWS Bedrock
fromBedrock
Custom / fromFetch
Raw HTTP escape hatch