Low-refusal inference API for authorized security work. OpenAI- and Anthropic-compatible.

Website · Docs · Benchmarks · GitHub

Featured: abliterated-benchmarks, updated every 8 hours.