Proxy Tutorial for OpenRouter (Free Models Only)

This tutorial shows you how to set up OpenRouter as a proxy in your chat app using free models only — no credit card or spend required.

Why OpenRouter?

- One API key for hundreds of models, including many completely free ones

- Good for agentic workflows, long context, and general chat

- Easy setup and reliable routing

Currently Available Free Models (as of July 2026)

Free models on OpenRouter change often as providers rotate capacity, so always double-check https://openrouter.ai/collections/free-models before committing to one. As of today, the strongest current options include:

| Model | Model ID | Context | Notes |

|---|---|---|---|

| Tencent Hy3 | tencent/hy3:free | 262K | Currently the most-used free model; strong at reasoning, agentic tasks, tool-calling |

| NVIDIA Nemotron 3 Ultra | nvidia/nemotron-3-ultra-550b-a55b:free | 1M | Frontier-class open reasoning/orchestration model, huge context |

| NVIDIA Nemotron 3 Super | nvidia/nemotron-3-super-120b-a12b:free | 1M | Strong all-rounder, efficient MoE architecture |

| Poolside Laguna M.1 | poolside/laguna-m.1:free | 256K | Flagship coding/agentic model |

| Cohere North Mini Code | cohere/north-mini-code:free | 256K | Lightweight, good for coding/terminal tasks |

| Google Gemma 4 31B | google/gemma-4-31b-it:free | 262K | Multimodal (text + image), 140+ languages |

| OpenAI gpt-oss-20b | openai/gpt-oss-20b:free | 131K | Open-weight, good for lower-latency use |

| NVIDIA Nemotron Nano 9B V2 | nvidia/nemotron-nano-9b-v2:free | 128K | Small, fast, configurable reasoning |

For the simplest setup, you can also just use openrouter/free as the model ID — it's a router that automatically picks a working free model for your request.

Note: Free models have rate limits and lower priority during peak times, and providers may use your inputs/outputs for training on their free tier (check individual model pages for specifics, e.g. Poolside's models note this explicitly).

Step-by-Step Setup

1. Open Your Chat App

Open your latest chat (or start a new one). Tap the menu icon (≣) in the top right → API Settings.

2. Go to the Proxy Tab

Switch to the Proxy tab → Proxy Configurations → Tap + NEW.

3. Fill in the Configuration

- Configuration Name: anything you like, e.g. "OpenRouter Free"

- Model Name: one of the model IDs above, e.g. tencent/hy3:free

- Proxy URL: https://openrouter.ai/api/v1/chat/completions

- API Key: see below

How to Get Your OpenRouter API Key

1. Go to https://openrouter.ai

2. Sign up or log in (Google, GitHub, or email)

3. Click your profile picture (top right) → Settings

4. Select API Keys

5. Click Create Key

6. Give it a name

7. (Optional) Set a $0 credit limit to guarantee you can never be charged

8. Click Create, then immediately copy the key (starts with sk-or-v1-...) — you won't be able to see it again

Paste the key into the API Key field in your app.

4. Tap Save Changes

5. Test the Configuration

Tap Test. You should see a "Valid API Key. Configuration works!" message.

6. Refresh the page (or restart the chat) and start using it.

FAQ / Troubleshooting

Error: "A network error occurred... Failed to fetch" or rate limit message.

- Double-check the API Key, Model Name, and Proxy URL for typos

- Confirm your API key hasn't been revoked

- Try refreshing the app or clearing cache

- Free models get rate-limited during peak times — wait a few minutes and retry, or switch to another free model from the table above

Model ID not found:

Free model IDs sometimes change or get deprecated as providers rotate capacity. Check https://openrouter.ai/collections/free-models for the current list before assuming your config is broken.