Gemma 4 31B is now free on OpenRouter and NVIDIA
Gemma 4 31B has free endpoints on OpenRouter and NVIDIA, with a 262K context window
Google DeepMind’s Gemma 4 31B now has free endpoints on OpenRouter and NVIDIA, with no charge for input or output. It offers a 262K context window and can handle text and images. On OpenRouter the model ID is google/gemma-4-31b-it:free; the free endpoint is rate-limited. NVIDIA’s free endpoint requires a signed-in account and an API key, at about 40 requests per minute.
- Eligible plans
- The OpenRouter free route, plus the NVIDIA Free Endpoint
- Eligibility
- An OpenRouter or NVIDIA account can call the matching free endpoint. NVIDIA requires a signed-in account and an API key. The free routes are rate-limited, and prompts and responses may be used to improve the models.
- Benefit duration
- Free for a limited time; no end date has been announced
The briefing
Google DeepMind’s Gemma 4 31B now has free endpoints on OpenRouter and NVIDIA, with no charge for input or output. It is the 31B dense model in the Gemma 4 family, with about 30.7 billion parameters. It handles text and images, can also read video as a sequence of frames, supports a switchable reasoning mode, and can call tools. The context window is 262K, the same 256K-token window listed on the model card. It was pretrained across 140+ languages and is released under Apache 2.0.
What you can use
The OpenRouter model ID is google/gemma-4-31b-it:free. The free endpoint allows up to 20 requests per minute. Accounts that have purchased fewer than 10 Credits get up to 50 requests per day; after at least 10 Credits have been purchased, the daily cap rises to 1,000. Completions are capped at about 32,000 tokens.
NVIDIA offers a Free Endpoint on build.nvidia.com. The model ID is google/gemma-4-31b-it. You need to sign in and create an API key. The free tier allows about 40 requests per minute and is meant for prototyping and testing. Your actual rate limit is shown in the console.
The Kilo catalog no longer lists google/gemma-4-31b-it:free. You can still use OpenRouter or NVIDIA. Neither of these two free routes has an announced end date. Prompts and responses on the free routes may be used to improve the models.
How to use
- Create an OpenRouter API key, set the base URL to
https://openrouter.ai/api/v1, and use the model namegoogle/gemma-4-31b-it:free. - Open the NVIDIA model page, sign in, join the developer program, create an API key, set the base URL to
https://integrate.api.nvidia.com/v1, and use the model namegoogle/gemma-4-31b-it.
See the Gemma 4 model card, the OpenRouter model page, the OpenRouter free-endpoint limits, and the NVIDIA model page.
