Hi,
I wanted to share this GitHub page I found, which gathers a collection of free and trial LLM APIs. "a list of free LLM inference resources accessible via API." "This lists various services that provide free access or credits towards API-based LLM usage."
If you're looking for free APIs, you'll find plenty here that can be used for translation. You might even check the GitHub page's forks, as there could be even more options available!
I've put together a small, dedicated guide for these APIs. I’ll explain how to set up each free API for GameTranslate, plus a few extras that aren’t in the list. If I messed something up or wasn’t 100% accurate, sorry about that. This guide is mainly for people who aren't particularly tech-savvy, users who aren't familiar with technical setups or who need a straightforward walkthrough to get things set up.
API Setup Guide
1. Openrouter: (e.g., xAI: Grok 4 Fast - free)
API Key: Click your profile (top right) -> settings -> Create API Key -> Name it -> Copy it now and save it somewhere safe, as you won't see it again. Done!
You can find loads of other free APIs by searching for 'free' under the models section: Link
Code:
Endpoint URL: https://openrouter.ai/api/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "x-ai/grok-4-fast:free",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
],
"stream": false,
"reasoning": {
"enabled": false
}
}
Text Output Path: .choices[0].message.content To switch models, you just change the model's name in the Body. Get the model name from the API tab on the model's page, under the curl example, e.g.: "model": "x-ai/grok-4-fast:free". Other examples are deepseek/deepseek-chat-v3.1:free or qwen/qwen3-235b-a22b:free. Find what works best for you!
Limits/Pricing: Free models are limited to 20 requests per minute. There's a 50 request per day limit if you haven't bought credits, or 1000 requests per day if you buy at least 10 credits. Link
API Key: Head to Projects (left menu) -> Click Create a new project (top right) -> Give it a name -> Create project -> Go to API keys (left menu) -> Click Create API Key (top right) -> Name it -> Select your new project -> Click Create Key. That should do it!
Code:
Endpoint URL: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash-lite:generateContent
Headers:
{
"Content-Type": "application/json",
"X-goog-api-key": "your api key"
}
Body:
{
"contents": [
{
"parts": [
{
"text": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
]
}
],
"generationConfig": {
"thinkingConfig": {
"thinkingBudget": 0
}
}
}
This already includes reasoning being disabled, so you can safely use reasoning models as well.
Text Output Path: candidates[0].content.parts[0].text
To switch to a different model, all you have to do is change the model's name directly in the Endpoint URL. Look for where it says models/ and replace the name right before :generateContent. For instance, you would change generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash-lite:generateContent to generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-lite:generateContent.
You can see the full list of models on the official documentation page or here: Google AI Studio -> Top right click on the model, where it says e.g., Gemini 2.5 Pro and you can see them there too. Crucially, remember to use the hyphenated version of the model name, like gemini-2.5-pro, not the standard name like Gemini 2.5 Pro.
Limits/Pricing: You can check the rate limits either on the GitHub page or on Google's official Free Tier documentation.
3. Nvidia NIM
API Key: You'll need to verify your phone number first. Look for the 'verify' link at the top when you sign up to enter your number. -> You can get the key from the top right: Get API Key or via your profile under API Keys -> Generate API Key -> Name it and set the Expiration (1 year max). -> Click Generate Key. Important: Make sure to save your API key right away, because "This is the only time your key will be displayed. This key is for API testing use only and is valid for 1 year."
Code:
Endpoint URL: https://integrate.api.nvidia.com/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
} Body:
{
"model": "qwen/qwen3-next-80b-a3b-instruct",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text% /no_think"
}
],
"stream": false
} Text Output Path: .choices[0].message.content
To switch models, you just change the model's name in the Body. Models: Check the documentation here. Select a model from the left, click to expand, and then find the model part in the code snippet on the right, for example: "model": "qwen/qwen3-coder-480b-a35b-instruct". Paste that name in.
Limits/Pricing: 40 requests per minute
API Key: In the left-side menu, go to API Keys -> Choose a Plan -> Select Experiment for free -> Click Subscribe -> You'll need to enter your phone number. -> Then, go back to API Keys -> Create new key -> Name your key and click Create new key again. You'll see this warning: "API key successfully created. Please copy it now, it will not be shown again. Note that it may take a few minutes to be usable."
Code:
Endpoint URL: https://api.mistral.ai/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "magistral-small-2509",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
],
"stream": false,
"prompt_mode": null
}
Text Output Path: .choices[0].message.content
To switch models, simply update the model's name in the Body section. You can find the available models on their documentation page.
Limits(per-model)/Pricing: 1 request/second, 500,000 tokens/minute, 1,000,000,000 tokens/month Link
This is on the same website as the previous one. Just find Codestral in the left menu -> Click Request Access -> Check the box and Accept and request access.
API Key: Under the Codestral section -> Click Generate API Key. You're all set.
Code:
Endpoint URL: https://codestral.mistral.ai/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "codestral-2508",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
],
"stream": false,
"prompt_mode": null
}
Text Output Path: .choices[0].message.content
To switch models, simply update the model's name in the Body section. You can find the available models on their documentation page.
Important: This API will only work with Codestral models, such as codestral-2508.
Limits(per-model)/Pricing: You're limited to 30 requests per minute, and 2,000 requests per day.
6. HuggingFace Inference Providers e.g: Nebius AI
Token Key: Go to your profile settings (top right) -> Access Tokens (left menu) -> Create new token -> Choose Read or Write access and name it -> Create Token. Remember this warning: "Save your token value somewhere safe. You will not be able to see it again after you close this modal. If you lose it, you'll have to create a new one."
Code:
Endpoint URL: https://router.huggingface.co/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "meta-llama/Llama-3.1-8B-Instruct:nebius",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
],
"stream": false
}
Text Output Path: .choices[0].message.content
If you want to use a different model, you just change the model name in the Body. Models: You can find the list of partners providing APIs here. A simpler view is on the models page under the Inference Providers filter. To find the model ID: Select a model (e.g., meta-llama / Llama-3.1-8B-Instruct), then click Deploy on the right -> Inference Providers -> Look for the model string, like: "model": "meta-llama/Llama-3.1-8B-Instruct:nebius". Copy that into your app. Limits/Pricing: You get $0.10/month in credits. See the pricing page for details.
API Key: You'll need to enter your credit card details during registration, but relax—it won't charge you anything unless you decide to upgrade. Like DeepL, it just stops working if you hit your limit. Steps: Profile (top right) -> Dashboard -> AI Gateway -> Create an API Key -> Create Key -> Name it and Create Key. Make sure to save the key because Vercel warns: "Save this key securely—it won't be shown again. Keep it safe, as anyone with access can make requests on your behalf."
Code:
Endpoint URL: https://ai-gateway.vercel.sh/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "xai/grok-4-fast-non-reasoning",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text%"
}
],
"stream": false
}
Text Output Path: .choices[0].message.content
To switch models, you simply change the model's name in the Body. You can browse all the available models here. Pick one, like xai/grok-4-fast-non-reasoning, and the correct model name will be right there for you to use.
Limits/Pricing: Link. Free Tier Details: Every Vercel team account gives you $5 of free usage per month to play around with the AI Gateway at no initial cost. Here’s how the free tier works: You get a $5 credit every 30 days after your first request. This credit works across their entire model catalog. You can stay on the free tier indefinitely as long as you don't buy extra credits. If you move to a paid tier: Once you purchase credits, your account switches to a pay-as-you-go model. You won't get the $5 monthly free credit anymore, but you'll have more capacity.
8. Cerebras
API Key: The key is created automatically when you register, and you can find it later in the left menu under the API keys tab.
Code:
Endpoint URL: https://api.cerebras.ai/v1/chat/completions
Headers:
{
"Content-Type": "application/json",
"Authorization": "Bearer your api key"
}
Body:
{
"model": "qwen-3-235b-a22b-instruct-2507",
"messages": [
{
"role": "user",
"content": "Your prompt goes here, for example Translate this text to English and only return the translated text: %text% /no_think"
}
],
"stream": false
}
Text Output Path: .content
To switch models, just change the model's name in the Body. Models: You'll need the Model ID which is listed here. If you click on the models, you'll see settings that let you disable "reasoning" (the model's internal thinking process). For example, Qwen 3 235B Instruct only supports "non-thinking mode," so you won't see any <think></think> tags.
For Qwen 3 32B, you can still disable the reasoning for speed by adding /no_think to your prompt (e.g., Tell me about cats /no_think). A heads-up though: Even with reasoning turned off, the empty <think></think> tags will still appear in your output. If you can live with the tags being there, the speed boost makes this a great option.
Limits/Pricing: You can find the specific limits on the GitHub page or by checking the details for each model on the official overview page.