Free LLM API comparison
Every free LLM API provider in our directory, side by side. Compare the type of free access, whether a credit card is needed, rate limits and headline models. Click a provider for full setup instructions and code.
| Provider | Access | No card | OpenAI-compat | Top models | Free limits |
|---|---|---|---|---|---|
| Free tier | Gemini 2.5 Flash, Gemini 2.5 Pro | Gemini 2.5 Flash ≈ 10 RPM / 250 req per day (free tier). | |||
| Free tier | Llama 3.3 70B, DeepSeek R1 Distill | Typically 30 RPM and 1,000–14,400 requests/day per model (free tier). | |||
| Free tier | DeepSeek V3 (free), Llama 4 Maverick (free) | ~50 free requests/day, or ~1,000/day after adding ≥ $10 lifetime credit. | |||
| Free tier | Llama 3.3 70B, Llama 4 Scout | Around 30 RPM and ~1M tokens/day on the free tier (varies by model). | |||
| Free tier | GPT-4o, GPT-4o mini | Low free rate limits (per-minute + per-day requests) tiered by model size. | |||
| Free tier | Mistral Small 3.2, Codestral | ~1 request/sec and a monthly token ceiling on the free plan. | |||
| Free tier | Llama 3.3 70B, Llama 4 Scout | 10,000 Neurons/day free (varies by model cost per request). | |||
| Free tier | Llama 3.3 70B, DeepSeek R1 | Per-model RPM limits on the free tier; generous for testing. | |||
| Free credits | Llama 3.3 70B, DeepSeek R1 | Shared free rate limits; credits refresh for developer accounts. | |||
| Free credits | Llama 3.3 70B Turbo (free), DeepSeek R1 | Free endpoint is rate-limited; credit covers the rest of the catalog. | |||
| Free tier | Qwen 3 Max, Qwen 3 Coder | Free quota is a fixed token grant per model (e.g. ~1M tokens), then paid. | |||
| Free credits | Llama 3.3 70B, DeepSeek V3 | Small monthly credit (larger for PRO); then pay-as-you-go. | |||
| Free credits | Llama 3.3 70B, DeepSeek V3 | Credit-based; no hard daily cap while credit lasts. | |||
| Free tier | GLM-4-Flash, GLM-4.6 | Free Flash models are rate-limited per minute; fair for app use. | |||
| Free tier | Command A, Command R+ | Trial keys: ~20 calls/min and a monthly cap; production needs a paid key. | |||
| Free credits | Llama 4 Maverick, DeepSeek V3 | Credit-based with developer rate limits; no hard daily free cap. | |||
| Low cost | DeepSeek V3.2, DeepSeek R1 | No hard rate cap; pay-as-you-go after the starter credit. | |||
| No key needed | OpenAI-compatible (text), FLUX (images) | Shared public rate limits; best for light / hobby use. | |||
| Free credits | Grok 4, Grok 4 Fast | Credit-based; standard per-minute limits on the console. | |||
| Free tier | Llama 3.3 70B, Mistral Small | Free tier capped by tokens-per-minute; EU region only. | |||
| Free tier | Llama 3.3 70B, Mistral | Beta free rate limits per model; EU-hosted. |
Google AI Studio (Gemini) free API →Groq free API →OpenRouter free API →Cerebras free API →GitHub Models free API →Mistral AI free API →Cloudflare Workers AI free API →SambaNova Cloud free API →NVIDIA NIM free API →Together AI free API →Alibaba Qwen (DashScope) free API →Hugging Face free API →Hyperbolic free API →Zhipu AI (GLM) free API →Cohere free API →Fireworks AI free API →DeepSeek free API →Pollinations AI free API →xAI (Grok) free API →Scaleway Generative APIs free API →OVHcloud AI Endpoints free API →