Free LLM API Resources Directory

Use the following summary of free providers to select a service based on your rate limit and model requirements:

OpenRouter

• Limits: 20 requests/minute, 50 requests/day (can increase to 1000 requests/day with a $10 lifetime top-up).
• Models: Includes Llama 3.1/3.2/3.3, Hermes 3, Google Gemma 4, NVIDIA Nemotron, and more.

Google AI Studio

• Limits: Varies by model. For example, Gemini 3.5 Flash allows 5 requests/minute and 20 requests/day. Gemma 3 models allow higher throughput (up to 30 requests/minute).
• Note: Data may be used for training if used outside of the UK/CH/EEA/EU.

NVIDIA NIM

• Limits: 40 requests/minute.
• Requirement: Phone number verification required.

Mistral

• La Plateforme: 1 request/second, 500,000 tokens/minute. Requires phone verification and opting into data training for the free tier.
• Codestral: 30 requests/minute, 2,000 requests/day. Requires phone verification.

Cerebras

• Limits: 30 requests/minute, 60,000 tokens/minute, 900 requests/hour.
• Models: gpt-oss-120b, Llama 3.1 8B.

Groq

• Limits: Varies significantly by model. Llama 3.3 70B is limited to 1,000 requests/day and 12,000 tokens/minute, while Llama 3.1 8B allows 14,400 requests/day.

Cohere

• Limits: 20 requests/minute, 1,000 requests/month.

GitHub Models

• Limits: Dependent on your GitHub Copilot subscription tier. Features extremely restrictive input/output token limits.

Cloudflare Workers AI

• Limits: 10,000 neurons/day.
• Models: Various Llama, Gemma, Mistral, and Qwen models.

The following providers offer one-time or limited-term trial credits to test their LLM models. Use these to evaluate model performance before committing to a paid plan.

• Alibaba Cloud (International) Model Studio: Provides 1 million tokens per model for various Qwen models.
• Modal: Offers $5/month upon sign up, increasing to $30/month if a payment method is added. Models are billed based on compute time.
• Inference.net: Provides $1 in credits, or $25 if you respond to an email survey.
• Hyperbolic: Provides $1 in credits. Supported models include DeepSeek V3, Llama 3.3 70B Instruct, deepseek-ai/deepseek-r1-0528, and qwen/qwen3-coder-480b-a35b-instruct.
• SambaNova Cloud: Provides $5 for 3 months. Supported models include deepseek-v3.1, deepseek-v3.2, gemma-4-31b-it, gpt-oss-120b, meta-llama-3.3-70b-instruct, and minimax-m2.7.
• Scaleway Generative APIs: Provides 1,000,000 free tokens. Supported models include a wide range such as Gemma 3 27B Instruct, Llama 3.3 70B Instruct, Pixtral 12B, Whisper Large v3, and various Qwen and Mistral models.