diff --git a/docs/inference-providers/_toctree.yml b/docs/inference-providers/_toctree.yml
index 4623dd474f..af2273fd37 100644
--- a/docs/inference-providers/_toctree.yml
+++ b/docs/inference-providers/_toctree.yml
@@ -36,8 +36,12 @@
sections:
- local: integrations/index
title: Overview
- - local: integrations/adding-integration
- title: Add Your Integration
+ - local: integrations/pi
+ title: Pi
+ - local: integrations/opencode
+ title: OpenCode
+ - local: integrations/codex
+ title: Codex
- local: integrations/claude-code
title: Claude Code
- local: integrations/hermes-agent
@@ -46,14 +50,12 @@
title: NeMo Data Designer
- local: integrations/macwhisper
title: MacWhisper
- - local: integrations/opencode
- title: OpenCode
- - local: integrations/pi
- title: Pi
- local: integrations/visionagents
title: Vision Agents
- local: integrations/vscode
title: VS Code with GitHub Copilot
+ - local: integrations/adding-integration
+ title: Add Your Integration
- local: tasks/index
title: Inference Tasks
diff --git a/docs/inference-providers/index.md b/docs/inference-providers/index.md
index 4572e7c280..9f65ac6575 100644
--- a/docs/inference-providers/index.md
+++ b/docs/inference-providers/index.md
@@ -7,6 +7,33 @@
Hugging Face’s Inference Providers give developers access to hundreds of machine learning models, powered by world-class inference providers. They are also integrated into our client SDKs (for JS and Python), making it easy to explore serverless inference of models on your favorite providers.
+## Quick Setup for Agents
+
+Using a coding agent? Point it at Inference Providers to run [the latest open models available](https://huggingface.co/models?inference_provider=all) with a single Hugging Face token. Pick your tool below to jump straight to its setup guide.
+
+
+
## Partners
Our platform integrates with leading AI infrastructure providers, giving you access to their specialized capabilities through a single, consistent API. Here's what each partner supports:
diff --git a/docs/inference-providers/integrations/claude-code.md b/docs/inference-providers/integrations/claude-code.md
index 998f1c18d1..f429b50069 100644
--- a/docs/inference-providers/integrations/claude-code.md
+++ b/docs/inference-providers/integrations/claude-code.md
@@ -1,6 +1,6 @@
# Claude Code
-[Claude Code](https://docs.anthropic.com/en/docs/claude-code) is Anthropic's agentic coding tool that lives in your terminal. It can understand your codebase, edit files, run commands, and handle complex development tasks. By pointing it at Hugging Face Inference Providers, you can use open models like [GLM-5.1](https://huggingface.co/zai-org/GLM-5.1), [Gemma 4](https://huggingface.co/google/gemma-4-31B-it), and more as the backing LLM.
+[Claude Code](https://docs.anthropic.com/en/docs/claude-code) is Anthropic's agentic coding tool that lives in your terminal. It can understand your codebase, edit files, run commands, and handle complex development tasks. By pointing it at Hugging Face Inference Providers, you can use any of [the latest open models available](https://huggingface.co/models?inference_provider=all) as your backing LLM.
## Overview
@@ -63,9 +63,9 @@ Then start Claude Code:
claude
```
-Replace `zai-org/GLM-5.1` with any model available on [Inference Providers](https://huggingface.co/inference/models). You can also append a provider suffix to pin a specific provider (e.g. `MiniMaxAI/MiniMax-M2.7:fireworks-ai`). Setting an explicit provider gives you deterministic routing, while omitting it (e.g. `MiniMaxAI/MiniMax-M2.7`) lets the router fall back to alternative providers for better resilience.
+Replace `zai-org/GLM-5.1` with any model available on [Inference Providers](https://huggingface.co/inference/models). You can also append a provider suffix to pin a specific provider (e.g. `MiniMaxAI/MiniMax-M2.7:fireworks-ai`). Setting an explicit provider gives you deterministic routing, while omitting it (e.g. `MiniMaxAI/MiniMax-M2.7`) uses the default `:fastest` policy. With this policy, the router picks the provider with the highest throughput, and automatically fails over if it becomes unavailable.
-You can also append a `:cheapest` or `:fastest` suffix to prefer cheaper or faster providers (e.g. `MiniMaxAI/MiniMax-M2.7:cheapest`).
+Available policies are `:cheapest`, `:fastest`, or `:preferred`; use one of these suffixes to prefer cheaper, faster, or your preferred-order providers (e.g. `MiniMaxAI/MiniMax-M2.7:cheapest`).
> [!TIP]
> The `ANTHROPIC_DEFAULT_*_MODEL` variables map to Claude Code's internal model slots (Opus, Sonnet, Haiku), from the most powerful to the quickest. You can assign different models to each slot to balance capability and speed e.g. `zai-org/GLM-5.1` for Opus, `google/gemma-4-31B-it:together` for Sonnet, and `openai/gpt-oss-120b:cerebras` for Haiku. `CLAUDE_CODE_SUBAGENT_MODEL` controls which model is used for sub-agents.
diff --git a/docs/inference-providers/integrations/codex.md b/docs/inference-providers/integrations/codex.md
new file mode 100644
index 0000000000..a1eb6f47ff
--- /dev/null
+++ b/docs/inference-providers/integrations/codex.md
@@ -0,0 +1,88 @@
+# Codex
+
+[Codex](https://developers.openai.com/codex) is OpenAI's agentic coding CLI that runs in your terminal, reads and edits your codebase, runs commands, and handles multi-step development tasks. By pointing it at Hugging Face Inference Providers, you can use any of [the latest open models available](https://huggingface.co/models?inference_provider=all) as your backing model.
+
+## Overview
+
+Codex lets you define custom model providers in its configuration file. By adding the Hugging Face router as a provider, all Codex requests are routed through Inference Providers, giving you access to a wide range of open models behind an OpenAI-compatible API.
+
+## Prerequisites
+
+- Codex CLI installed ([installation guide](https://developers.openai.com/codex))
+- A Hugging Face account with [API token](https://huggingface.co/settings/tokens/new?ownUserPermissions=inference.serverless.write&tokenType=fineGrained) (needs "Make calls to Inference Providers" permission)
+
+## Configuration
+
+### Quick Setup
+
+1. Create a Hugging Face token with Inference Providers permissions at [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens/new?ownUserPermissions=inference.serverless.write&tokenType=fineGrained)
+
+2. Set your token as an environment variable:
+
+```bash
+export HF_TOKEN=hf_...
+```
+
+3. Add Hugging Face as a model provider in `~/.codex/config.toml`:
+
+```toml
+[model_providers.huggingface]
+name = "Hugging Face"
+base_url = "https://router.huggingface.co/v1"
+env_key = "HF_TOKEN"
+wire_api = "responses"
+```
+
+4. Create a profile in its own file at `~/.codex/huggingface.config.toml` that selects this provider and a model:
+
+```toml
+model_provider = "huggingface"
+model = "openai/gpt-oss-120b"
+```
+
+> [!NOTE]
+> For Codex versions before 0.134.0, you have to define profiles in a `[profiles.]` table inside `config.toml`.
+
+5. Launch Codex with the Hugging Face profile:
+
+```bash
+codex --profile huggingface
+```
+
+You can also run a one-off command non-interactively:
+
+```bash
+codex exec --profile huggingface "Explain what this repository does."
+```
+
+Replace `openai/gpt-oss-120b` with any model available on [Inference Providers](https://huggingface.co/models?other=conversational&inference_provider=all&sort=trending). You can append a provider suffix to pin a specific provider (e.g. `openai/gpt-oss-120b:groq`). Setting an explicit provider gives you deterministic routing, while omitting it lets the router fall back to alternative providers for better resilience. You can also append a `:cheapest` or `:fastest` suffix to prefer cheaper or faster providers.
+
+To override the model for a single run without editing your config, pass `-m`:
+
+```bash
+codex exec --profile huggingface -m "zai-org/GLM-5.1" "Return OK."
+```
+
+> [!TIP]
+> Codex custom providers require `wire_api = "responses"`, which routes requests through the OpenAI-compatible [Responses API](../guides/responses-api). Make sure the model you select is available for chat completion on Inference Providers.
+
+### Billing to an Organization
+
+To bill inference usage to a Hugging Face organization instead of your personal account, add the `X-HF-Bill-To` header to the provider definition in `~/.codex/config.toml`:
+
+```toml
+[model_providers.huggingface]
+name = "Hugging Face"
+base_url = "https://router.huggingface.co/v1"
+env_key = "HF_TOKEN"
+wire_api = "responses"
+http_headers = { "X-HF-Bill-To" = "your-org-name" }
+```
+
+Replace `"your-org-name"` with the name of the organization you want to bill to. You must have Write privileges in the org.
+
+## Resources
+
+- [Codex Documentation](https://developers.openai.com/codex)
+- [Codex Advanced Configuration](https://developers.openai.com/codex/config-advanced)
+- [Available models on Inference Providers](https://huggingface.co/models?other=conversational&inference_provider=all&sort=trending)
diff --git a/docs/inference-providers/integrations/index.md b/docs/inference-providers/integrations/index.md
index 93afe264cb..8fcfdd400b 100644
--- a/docs/inference-providers/integrations/index.md
+++ b/docs/inference-providers/integrations/index.md
@@ -18,6 +18,7 @@ This table lists _some_ tools, libraries, and applications that work with Huggin
| Integration | Description | Resources |
| ------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- |
| [Claude Code](https://docs.anthropic.com/en/docs/claude-code) | Agentic coding tool for the terminal | [HF docs](./claude-code) |
+| [Codex](https://developers.openai.com/codex) | OpenAI's agentic coding CLI for the terminal | [HF docs](./codex) |
| [CrewAI](https://www.crewai.com/) | Framework for orchestrating AI agent teams | [Official docs](https://docs.crewai.com/en/concepts/llms#hugging-face) |
| [NeMo Data Designer](https://github.com/NVIDIA-NeMo/DataDesigner) | Synthetic dataset generation framework | [HF docs](./datadesigner) |
| [GitHub Copilot Chat](https://docs.github.com/en/copilot) | AI pair programmer in VS Code | [HF docs](./vscode) |
@@ -55,6 +56,7 @@ End-user applications and interfaces powered by LLMs.
AI-powered coding assistants and development environments.
- [Claude Code](https://docs.anthropic.com/en/docs/claude-code) - Agentic coding tool for the terminal ([HF docs](./claude-code))
+- [Codex](https://developers.openai.com/codex) - OpenAI's agentic coding CLI for the terminal ([HF docs](./codex))
- [GitHub Copilot Chat](https://docs.github.com/en/copilot) - AI pair programmer in VS Code ([HF docs](./vscode))
- [Hermes Agent](https://github.com/NousResearch/hermes-agent) - Open-source AI agent CLI for coding and development ([Official docs](https://hermes-agent.nousresearch.com/docs/user-guide/configuration/#hugging-face-inference-providers) / [HF docs](./hermes-agent))
- [OpenCode](https://opencode.ai/) - AI coding agent built for the terminal ([Official docs](https://opencode.ai/docs/providers#hugging-face) / [HF docs](./opencode))
diff --git a/docs/inference-providers/integrations/pi.md b/docs/inference-providers/integrations/pi.md
index e5e26423fb..c40200647f 100644
--- a/docs/inference-providers/integrations/pi.md
+++ b/docs/inference-providers/integrations/pi.md
@@ -8,7 +8,7 @@ Pi natively supports Hugging Face Inference Providers, giving you access to open
## Prerequisites
-- Pi installed (`npm install -g @mariozechner/pi-coding-agent`)
+- Pi installed (`curl -fsSL https://pi.dev/install.sh | sh`)
- A Hugging Face account with [API token](https://huggingface.co/settings/tokens/new?ownUserPermissions=inference.serverless.write&tokenType=fineGrained) (needs "Make calls to Inference Providers" permission)
## Configuration
@@ -43,7 +43,7 @@ Alternatively, store it in `~/.pi/agent/auth.json`:
}
```
-4. Start pi and run the `/model` command (or press **Ctrl+L**) to open the interactive model selector. You can also press **Ctrl+P** to cycle through your scoped models.
+4. Start pi and run the `/model` command (or press **Ctrl+L**) to open the interactive model selector. You can also press **Ctrl+P** to cycle to the next model.
### Billing to an Organization
@@ -65,5 +65,5 @@ Replace `"your-org-name"` with the name of the organization you want to bill to.
## Resources
-- [Pi Documentation](https://github.com/badlogic/pi-mono)
+- [Pi Documentation](https://pi.dev/docs/latest)
- [Available models on Inference Providers](https://huggingface.co/models?pipeline_tag=text-generation&inference_provider=all&sort=trending)