Hyperbolic Free Inference Tier

Overview

Hyperbolic is an open-access AI cloud whose Serverless Inference service is exposed through an OpenAI-compatible API at https://api.hyperbolic.xyz/v1. The Basic tier has a $0 minimum deposit and provides free API access to the serverless model catalogue at 60 requests per minute.

Free Tier Quotas & Limits

• Basic tier: 60 RPM, $0 minimum deposit, API access to all models, community support.
• Pro tier: 600 RPM, priority queue and email support, requiring a deposit of at least $5.
• Each source IP is capped at 600 RPM for DDoS protection.
• Compute credits are purchased separately (minimum $5), are 1:1 with the dollar amount, and never expire. A positive balance is required for GPU instances, while inference services remain available to free-plan users.
• Zero data retention is advertised for API calls.

Key Free Models

The catalogue covers 25+ open-source models, including Llama 3.1 (8B, 70B, 405B), Qwen 2.5 (7B, 72B), DeepSeek V2.5, Hermes 3, Mistral 7B, Stable Diffusion XL/3.5, FLUX.1 [schnell/dev] and vision-language models such as Llama 3.2 Vision and Qwen2-VL. Model IDs follow provider paths such as meta-llama/Meta-Llama-3.1-70B-Instruct. GPT-OSS 120B/20B, the Llama 3.1 405B base model, FLUX.1-dev, the SDXL variants and Melo TTS are listed as scheduled for deprecation.

How to Get Started

1. Sign up at app.hyperbolic.ai and generate an API key.
2. Point any OpenAI SDK client at the base URL https://api.hyperbolic.xyz/v1 and use your key.
3. Send requests at up to 60 RPM on the free Basic tier.