Breaking

GitHub Repo Cataloging Permanently Free LLM APIs Gains Traction

July 4, 2026 at 02:45 EDT

Open Source
Foundation Models
Software Dev & Coding

A GitHub repository, "mnfst/awesome-free-llm-apis," that catalogs LLM APIs with permanent free tiers is spreading among developers and drawing attention. Unlike time-limited free trials, it focuses on providers offering permanently available free access with no credit card required, laying out model names, context length, maximum output, modality, and rate limits in table form.

June 2026 · awesome-free-llm-apis

A living list of LLMs that are permanently free — no trial, no card, no clock

Against rising 2025–2026 API costs and "freemium fatigue," a GitHub repository curates only permanent free tiers — spelling out rate limits, speed, and free-model scope. Most endpoints are OpenAI SDK-compatible, so they drop straight into Cursor, Claude Code, or Aider.

Strict inclusion rules: no trial credits, no time-limited promos, no credit card

Categories: Provider APIs & third-party Inference providers

Source of truth: data.json, README auto-generated via GitHub Actions

Inference speed, drawn to scale

Approx. tokens per second on free tiers — Cerebras runs ~87× faster than Groq's throughput class

~2,600

Cerebras

~30 RPM

Groq LPU

Cerebras bar ≈ 2,600 tok/s; Groq shown at its 30 RPM class — different metrics, both fast free tiers.

Selected permanent free tiers

Google Gemini PROVIDER

3.5 Flash — 15 RPM / 1,500 RPD

Mistral AI PROVIDER

~1B tokens / month (Experiment)

Cohere PROVIDER

1,000 calls / month (non-commercial)

Groq INFERENCE

llama-3.3-70b — 30 RPM / 1,000 RPD

Cerebras INFERENCE

1M tokens / day, ~2,600 tok/s

GitHub Models INFERENCE

45+ models — 10-15 RPM / 50-150 RPD

Cloudflare Workers AI INFERENCE

10,000 Neurons / day, 50+ models

Hugging Face INFERENCE

100K credits / month

Why developers like it

Clear, concrete rate limits — not vague "free AI" claims
OpenAI SDK-compatible: swap base URL, key, model
Great for side projects, prototyping, cost cutting
Free tiers can be stacked across services

The catches

Strict rate / daily request caps
Unavailable in some regions (e.g. the EU)
Prompts may be used for service improvement
Listings change often — verify the latest info

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.