AI & Machine Learning

Posted on Apr 25, 2026

Expand your applications with AI. This category covers generous free tiers and developer access for Large Language Models (LLMs), high-speed inference, vector databases for Retrieval-Augmented Generation (RAG), and ML hosting.

All figures and plan details below are as of 2026-04.

LLM & AI Inference APIs

  • Google Gemini API (AI Studio): Generous permanent free tier for Gemini 1.5 Flash and Pro. Includes 15 requests per minute (RPM), 1 million tokens per minute (TPM), and 1,500 requests per day (RPD). Google AI Studio
  • Groq: Blazing fast LPU inference for open models (Llama 3, Mixtral, Gemma). Offers a free tier with daily/minute rate limits (e.g., 14,400 requests/day for Llama 3 8B). GroqCloud
  • OpenRouter: AI API aggregator that maintains a list of permanently free models (like Llama, Mistral, and Gemma variants) routed through various providers. No credit card required for free models. OpenRouter Free Models
  • Cohere: Free developer tier for their Command (LLM) and Embed (embedding) models. Rate limited to 1,000 calls/month for prototyping and non-commercial use. Cohere Pricing
  • Hugging Face Serverless Inference: Free, rate-limited API to test and run thousands of open-source models hosted on the Hugging Face Hub. Hugging Face Inference API
  • Mistral AI (La Plateforme): Offers a free “Experiment” tier to test their open-weights models (Mistral Nemo, Codestral, etc.) via API. Mistral Pricing
  • GitHub Models: Free inference API access for developers to test top models (GPT-4o, Llama 3, Phi-3) directly through GitHub with daily rate limits. GitHub Models

Vector Databases (For RAG)

  • Pinecone: 1 free “Starter” index with up to 2GB of storage. Perfect for learning and small RAG projects. Pinecone Pricing
  • Qdrant: Free forever cluster on Qdrant Cloud (1GB RAM, 0.5 CPU, ~1M vectors). Qdrant Cloud Pricing
  • Weaviate: Free sandbox cluster that lasts for 14 days (renewable/re-deployable) for testing. Weaviate Cloud Pricing
  • Chroma: Open-source and can be run locally for free, but they also offer generous developer access for their hosted version. Chroma Pricing

ML Hosting, Notebooks & GPU

  • Hugging Face Spaces: Host ML demos and Gradio/Streamlit web apps for free on CPU (with 16GB RAM). Hugging Face Spaces
  • Google Colab: Free access to Jupyter notebooks backed by computing resources, including free (but transient and variable) access to T4 GPUs. Google Colab
  • Kaggle Kernels: Free Jupyter notebook environments offering up to 30 hours of free GPU (P100/T4) and 20 hours of TPU per week. Kaggle
  • Lightning AI: Free credits for Lightning Studios to build, train, and deploy AI products on cloud GPUs. Lightning AI

Free Trial Credits (Time-Limited)

  • OpenAI: New API accounts typically receive a small pool of free trial credits (valid for a few months). OpenAI API
  • Anthropic (Claude): Occasional free usage credits via the developer console for new accounts. Anthropic Console
  • Together AI: $5 in free credits for new users to explore fast model inference. Together AI Pricing