One API for every model, from every provider.

Heimdall is an inference marketplace. Providers list what they serve and set their own prices; you point an OpenAI-compatible client at one base URL and reach all of them.

Uptime, time to first token, throughput and public benchmark scores are measured by us on real traffic — never taken from a provider's own claims.

• Pay per token, at the provider's listed price
• Deposit in USDC, USDT or DAI across seven EVM networks
• Chat, vision and embeddings models