Provider

GitHub Models free API: what happened and what to use now

GitHub Models gave developers free, rate-limited access to many LLMs with a GitHub token. GitHub retired the service on 30 July 2026. This guide explains what stopped working, the replacements that GitHub recommends, and how to move code to free model tiers.

Status: GitHub Models is retired

GitHub announced the retirement in June 2026 and closed the service to new customers. On 1 July 2026, GitHub set the final date. The service stopped on 30 July 2026.

  • What stopped: the playground, the model catalog, the inference API and bring your own key (BYOK) endpoints.
  • Who it affects: all customers. This includes customers with active usage.
  • Before the end date: GitHub ran two short brownouts on 16 July and 23 July 2026. During each brownout, requests returned errors.
  • GitHub Spark: the Spark llm() function used GitHub Models. It stopped working on the same date.
Important: the free GitHub Models rate limits (requests per minute, requests per day and tokens per request) no longer apply to anything. Any code that calls the GitHub Models inference API now fails. Do not wait for the service to return.

The replacements that GitHub recommends

The GitHub changelog names two replacements. Neither one is a direct free copy of GitHub Models.

OptionWhat it isFree use
Microsoft Foundry (ai.azure.com)A model catalog and inference service on Azure, for new and existing projectsBilled through an Azure subscription. Check Azure for current trial offers
GitHub CopilotModels inside GitHub workflows, the editor and the Copilot agentCopilot plans have their own limits. It is not a general inference API for your app
Free provider tiersDirect free tiers from providers such as Google AI Studio, Groq, NVIDIA NIM and OVHYes, with rate limits per provider
FreeLLMAPIA self-hosted gateway that pools free tiers behind one OpenAI-compatible keyYes. The router is open source and calls only free tiers

If you used GitHub Models for prototypes, tests or a small side project, a free provider tier is the closest match. GitHub Models also used an OpenAI-style chat format. Most code moves with a change of base URL, key and model name.

Move your code to a free OpenAI-compatible endpoint

FreeLLMAPI runs on your computer. It sends each request to a free provider tier. When one provider returns a rate limit error, it tries the next provider. Do these steps:

  1. Install FreeLLMAPI with the one-line installer. Docker must be installed.
  2. Open the dashboard at http://localhost:3001.
  3. Add free keys from two or more providers on the Keys page. Google AI Studio and Groq each give a free key without a card.
  4. Copy your unified key from the Keys page.
  5. Change the base URL, the key and the model name in your code.

Install (macOS or Linux)

curl -fsSL https://freellmapi.co/install.sh | bash

Python: replace the GitHub Models client

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:3001/v1",
    api_key="<unified-key>",
)

resp = client.chat.completions.create(
    model="auto",  # or a model id from your catalog
    messages=[{"role": "user", "content": "Say hello in one line."}],
)
print(resp.choices[0].message.content)

Use auto to let the router pick a model. To keep one model, use a model id from the table below. Each response has an X-Routed-Via header. It shows the provider and the model that answered.

Free models that replace the GitHub Models catalog

GitHub Models listed OpenAI, Meta, Mistral, DeepSeek and Microsoft models. Many of the same model families have free tiers elsewhere. The table shows the strongest free models with tool calling and 128K or more context, from today's catalog.

Free models with tool calling and 128K+ context
ModelFree onContextFree limits (best offer)ToolsVision
GLM 5.3UnoRouter1M1 req/min
1 req/min per model
YesNo
GLM 5.3 FlashNVIDIA NIM, UnoRouter1M40 req/min
40 RPM
YesYes
Gemini 3.8 FlashGoogle AI Studio1MRecurring provider quotaYesYes
Qwen3.8 Flash NextAMD Radeon Cloud262KRecurring daily account allowanceYesNo
Qwen3.8 FlashElectronHub1M$0.25/week shared recurring creditYesYes
Gemini 3.7 FlashGoogle AI Studio1M10 req/min
20 req/day
250,000 tokens/min
~3M
YesYes
GPT-5.6 LunaElectronHub1M$0.25/week shared recurring creditYesYes
Dots3 Note PreviewKilo Gateway, OpenRouter512K20 req/min
50 req/day
50 req/day account-wide (1,000 after a $10 purchase)
YesYes
DeepSeek V4 FlashAMD Radeon Cloud, BazaarLink, ElectronHub1M10 req/min
50 req/day
Shared rate limit
YesNo
Qwen3.8 27BElectronHub, Groq, OVH AI Endpoints262K30 req/min
250 req/day
70,000 tokens/min
~6M
YesYes
Gemini 3.6 FlashGoogle AI Studio1M10 req/min
20 req/day
250,000 tokens/min
~3M
YesYes
Qwen 3.8 27BCloudflare Workers AI262KShared 10k neurons/dayYesYes

Showing 12 of 109 matching free models. See all models.

Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.

If your code used GPT models on GitHub Models, read Is there a free GPT API?. It lists the GPT-OSS open-weight models and other GPT routes with free access.

Differences from GitHub Models

  • Authentication: GitHub Models used a GitHub token. Free provider tiers use their own API keys. FreeLLMAPI gives you one unified key for all of them.
  • Limits: each provider sets its own limits. Some count requests per minute. Others count tokens per day or credits per month.
  • Model names: GitHub Models used names such as openai/gpt-4.1. Free providers use their own model ids. Check the catalog for the exact id.
  • Errors: a free tier returns error 429 when you reach its limit. Without a gateway, your code must catch this error and wait. FreeLLMAPI catches it and sends the request to the next provider.
  • Data use: some free tiers can use your prompts to improve their models. Read which free LLM APIs train on your data before you send private data.

Today the catalog lists 487 free models from 45 providers (checked 19 November 2026). One free tier stops at its limit fast. Add keys from four or more providers to get more requests per day.

Use every one of these through one key. FreeLLMAPI is an open-source, self-hosted router that puts all these free tiers behind a single OpenAI-compatible endpoint and fails over when one is rate-limited. Browse the catalog or go live.

Frequently asked questions

Is GitHub Models still free?

No. GitHub retired GitHub Models on 30 July 2026. The playground, the model catalog, the inference API and BYOK are not available to any customer.

What are the GitHub Models free rate limits?

They do not apply any more. The service is retired, so the old free limits per minute, per day and per request have no effect. Requests to the old inference API fail.

What does GitHub recommend instead of GitHub Models?

GitHub recommends Microsoft Foundry on Azure for model access in projects, and GitHub Copilot for AI workflows on GitHub. Neither one is a free general-purpose API.

How do I get a free replacement for GitHub Models?

Use a free provider tier such as Google AI Studio, Groq or NVIDIA NIM. To use several free tiers with one key and automatic failover, run FreeLLMAPI and point your OpenAI client at it.

Can I keep my OpenAI SDK code?

Yes. GitHub Models used an OpenAI-style chat format. Change the base URL, the API key and the model name. The rest of the code stays the same.

Sources

All guides · Browse all free models

Keep reading

All posts →
Provider

Free Gemini API

Google AI Studio's permanent free tier — Gemini and Gemma, no credit card.

Provider

Free DeepSeek API

DeepSeek V3, R1 and V4 served free across multiple providers.

Provider

Free Llama API

Meta's Llama models free across 14 providers, behind one key.