FreeTheAI
https://freetheai.org/models?q=Free
Free the ai offers multiple free AI models like Nemotron Ultra, GPT-4.1, Gemma, Moonshot Kimi, etc. They have context limits, with zero input/output token costs.
Open router
https://openrouter.ai/models?variant=free
This is a famous one that provides multiple Nemotron models. Since it updates regularly, the available models can change. Right now, they are providing 20+ models with free input and output tokens, subject to a context limit.
Google AI Studio
As everyone already knows, it has a very generous limit. Personally, I prefer using it for personal agentic coding and with team.
Requesty
https://www.requesty.ai/models?free=1
This one gives 200 requests/day, which is hmmm.. okay for personal use. It provides Nemotron, Laguna, Gemma, and Meta AI models for free.
NVIDIA NIM
https://build.nvidia.com/settings/api-keys
NVIDIA NIM on build.nvidia.com provides roughly 80+ free AI models at 40 requests/minute, including models like DeepSeek V4 Flash, Nemotron 3 Ultra, Laguna XS, Kimi K3, and MiniMax M3.
LLM7.io
llm7.io
It provides 3–5 free models like gpt-oss, Gemma 4 31B, MiniMax M2.7, Codestral, and Mistral Nemo. Getting a free token from token.llm7.io lifts the rate limit to **40 requests a minute **and unlocks more model IDs—fine for a first test.
ModelScope
https://modelscope.cn/my/myaccesstoken
ModelScope is Alibaba’s public model hub. The inference API offers a large free catalog with 45+ models on the live list—including MiniMax M3, Qwen 3.8, DeepSeek V4, and GLM—at 2,000 requests per 24 hours. It is OpenAI-compatible and has a max context of 1 million tokens.
OVHcloud
European hosting offers seven models at €0 in the catalog, while the rest can still be called without an account at 2 requests a minute. It provides models like NVIDIA Riva TTS (EN/DE/ES/IT), Qwen3Guard-Gen 0.6B and 8B, and Stable Diffusion XL.
If you Found This post helpful, so you can upvote this post.
Note : These providers can change their model lists or make free models paid at any time, so be sure to check their terms before using.
Top comments (0)