The Axiom

APIs

6 pages

Amazon Bedrock

Amazon Bedrock is AWS's managed foundation model service. The Converse API provides a unified boto3 interface across Claude, Nova, Llama, Mistral — same call, any model. Knowledge Bases add managed RAG. Guardrails add model-agnostic content safety.

awsbedrockconverse-apiknowledge-bases

Anthropic API

The Anthropic Messages API covers prompt caching, batch processing, streaming, tool use, and extended thinking — the features that matter most for production cost and quality optimisation.

anthropicapiclaudeprompt-caching

Google AI API (Gemini)

Google's Gemini API covers both Google AI Studio (developer) and Vertex AI (enterprise GCP) entry points, with the largest context window of any commercial model and competitive pricing for high-volume workloads.

googlegeminivertex-aigoogle-ai-studio

OpenAI API

The OpenAI API is the most widely integrated LLM API — nearly every framework supports it, many providers expose compatible endpoints, and it covers chat, function calling, embeddings, vision, audio, and reasoning models (o1/o3).

openaiapigptfunction-calling

OpenAI Responses API

OpenAI's stateful successor to Chat Completions (March 2025). Server-side conversation state via previous_response_id, built-in tools (web search, file search, code interpreter, MCP), and 40–80% better cache utilisation. Assistants API deprecated August 2025.

openairesponses-apiagentsstateful

What is an API?

An API is a contract that lets two pieces of software talk to each other without knowing each other's internals — you send a structured request, you get a structured response back.

apihttprestjson