TrueFoundry Models

A comprehensive, community-maintained registry of AI/LLM model configurations. This repository provides standardized model metadata including pricing, features, and token limits across all major AI providers.

Scope

This registry is metadata only, for providers the TrueFoundry LLM Gateway already integrates with. Every directory under providers/ corresponds to an existing gateway integration, and the list of those integrations lives in the gateway, not here.

That means:

• Adding or updating a model under an existing provider is exactly what this repo is for. Pricing, limits, features, modalities, deprecations — open a PR.
• Adding a new provider is not. A new providers/<name>/ directory has no effect on its own: without a gateway integration behind it, nothing reads those files at runtime. New provider support is gateway work, so open an issue describing the provider and the use case instead of sending registry YAML.

See Supported Providers below for the current list.

Why Use This?

LLM model configs change often — prices drop, features expand, limits shift. This repository provides up-to-date information across providers and makes updating stale data easy.

• Unified Schema — Consistent model configuration format across 24 providers
• Up-to-Date Pricing — Current cost information for input/output tokens, batch processing, and caching
• Feature Tracking — Know exactly what each model supports (vision, tools, structured output, etc.)
• Open Source — Community-driven updates ensure accuracy and coverage

Supported Providers

This is the complete set of providers the gateway supports. If a provider isn't listed here, the registry has no place to put its models yet. The table is generated from providers/ by update-readme-counts.ts, so don't edit it by hand.

Provider	Models	Description
OpenRouter	930	Unified API for open source models
Google Vertex AI	431	Gemini, PaLM on GCP
Together AI	384	Open source model hosting
Microsoft Foundry	299	OpenAI and Foundry catalog models on Microsoft Foundry
DeepInfra	262	Open source model hosting
Azure OpenAI	230	OpenAI models on Azure
AWS Bedrock	229	Claude, Llama, Titan, Mistral on AWS
Azure AI Foundry	220	Azure AI models
OpenAI	151	GPT-4, GPT-4o, GPT-5, o1, o3, DALL-E, Whisper, TTS
Deepgram	143	Speech-to-text and text-to-speech models
Mistral AI	103	Mistral, Mixtral, Codestral
xAI	92	Grok models
Google Gemini	80	Gemini Pro, Ultra, Flash
Databricks	68	Databricks-hosted models
Aws Bedrock Mantle	55
Cohere	39	Command, Embed models
SambaNova	30	Enterprise AI models
Anthropic	27	Claude 3, Claude 3.5, Claude 4
Perplexity	24	Search-augmented models
Groq	23	Fast inference models
Wafer	17
AI21	12	Jamba models
ElevenLabs	11	Voice synthesis and text-to-speech models
Cerebras	6	Fast inference models

Installation

Direct Clone

git clone https://github.com/truefoundry/models.git

Model Configuration Schema

Each model YAML file follows this schema (validated by CUE):

# Required
model: gpt-5.4-mini-2026-03-17        # Model identifier used by the provider's API
mode: chat                             # Primary capability (chat, embedding, image, text_to_speech, etc.)

# Pricing — array of cost entries, each with a region ("*" for global)
costs:
    - region: "*"
      input_cost_per_token: 7.5e-7
      output_cost_per_token: 0.0000045
      cache_read_input_token_cost: 7.5e-8

# Token and context window limits
limits:
    context_window: 400000
    max_output_tokens: 128000

# Feature flags
features: [function_calling, prompt_caching, structured_output, system_messages]

# Input/output modalities
modalities:
    input: [text, image]
    output: [text]

# Extended thinking / reasoning support
thinking: true

# Documentation or pricing source URLs
sources:
    - https://developers.openai.com/api/docs/pricing

See the Contributing Guide for the full list of fields, valid values, and more examples.

Directory Structure

providers/
├── <provider>/
│   ├── default.yaml        # Default params for all models under this provider
│   ├── <model>.yaml
│   └── ...

Example:

providers/
├── openai/
│   ├── default.yaml
│   ├── gpt-4o.yaml
│   ├── gpt-4o-mini.yaml
│   └── ...
├── anthropic/
│   ├── default.yaml
│   ├── claude-3-5-sonnet.yaml
│   └── ...
└── ...

Contributing

We welcome contributions! Whether it's adding a model under one of the supported providers, updating pricing, or fixing outdated information — see the Contributing Guide for details on the schema, examples, and PR process. For a provider that isn't on that list, read Scope first.

License

This project is licensed under the MIT License - see the LICENSE file for details.