Skip to main content
Deprecated: The Completions V1 API has been deprecated in favor of Completions v2.Key differences from V1:
  • New endpoint: /ai/v2/guarded/chat/completions
  • Model selection options: model, model_family, or optional auto_routing
  • Provider selection with model_family
  • Theology-aware responses with tradition
View V2 Guide | Migration Steps
Tool calling works in both V1 and V2, but the Tool Use Guide covers V2. Migrate to V2 for the latest tool calling features and examples. Standard chat completions, like OpenAI’s chat/completions endpoint, are the industry default for building conversational AI. Here’s why the standard makes sense for chatbots, virtual assistants and AI-powered automation.
The Gloo AI Completions API lets you build with many leading open-source models, with Human Flourishing alignment and guardrails for any use.

Why Use Standard Chat Completions?

Familiar Developer Experience

If you’ve used tools like OpenAI, Anthropic, or other LLM providers, you’ve likely already encountered the chat format: a sequence of messages between system, user, and assistant.
This message-based interface is the standard across APIs, so you can build, test, and scale without learning a custom format for every provider.

Plug-and-Play with Tools and SDKs

Standard chat completions work out of the box with:
  • Popular SDKs (OpenAI, LangChain, LlamaIndex, etc.)
  • Prompt engineering tools
  • Debugging/observability dashboards
  • Prompt versioning platforms
  • Orchestration frameworks
That means less boilerplate.

Easily Portable Across Models

Using a standard chat format lets you switch between:
  • OpenAI GPT and o-series
  • Anthropic Claude
  • Google Gemini
  • Open-source chat models like DeepSeek, LLaMA, or Mistral
…without rewriting your app logic. You can also benchmark models, compare outputs, and build model-agnostic fallbacks or ensembles.

Better Alignment with Human Intent

The chat format encourages natural interaction flow, including:
  • Clarification (multi-turn)
  • Function/tool calling
  • Role-based prompting (e.g. system as context setter)
It also makes features like memory, tool use, and retrieval-augmented generation (RAG) feel native, rather than bolted on.

Built-In Support for Advanced Features

Using chat completions gives you access to:

Supported by the Ecosystem

From hosting providers to vector databases, the AI tooling ecosystem expects the standard chat format, so using it keeps your codebase compatible.

TL;DR: Why Use Standard Chat Completions?

Next steps

Completions V2

The current Completions API, with model selection and routing.

Supported Models

Model IDs, capabilities, and pricing.

Completions V1 API Reference

Every request and response field for the V1 endpoint.

Completions V1 Tutorial

The step-by-step V1 build, kept for existing integrations.
The models V1 accepts are listed in Supported Models (Legacy).