Blackbox AI Launches Enterprise Inference Platform with 300+ Model Router and Dedicated Single-Tenant Capacity
Blackbox AI introduced single-tenant isolated Enterprise Inference along with the Blackbox Router, connecting over 300 open and closed models through a single OpenAI-compatible endpoint. Independent verification by Artificial Analysis rated Blackbox as the #1 Nemotron 3 Ultra provider at 454 tokens/sec, alongside announcements for Nemotron 3.5 Lightning achieving 1,200 tokens/sec.
Verified State Diff
Impact & Verification Analysis
Enterprise software engineering teams, AI infrastructure engineers, and DevOps administrators building agentic workflows and deploying open-weight models.
Provides high-throughput inference with strict privacy guarantees (ZDR) and dedicated hardware isolation, allowing enterprise security teams to adopt open and closed models without data leakage risks.
Full Fact Overview
Blackbox AI has updated its core developer platform to offer four distinct products: Enterprise Inference (dedicated open-weight deployments with zero data retention and PII stripping), Blackbox Router (a single gateway supporting 300+ models with provider fallback and load balancing), Agents API (HTTP execution of sandboxed coding agents with PR creation and log streaming), and a Command Line agentic terminal. The infrastructure enforces Zero Data Retention (ZDR) and end-to-end encryption across single-tenant isolated instances.