From the source
Baseten is launching Grid, a single layer for enterprises to route requests, monitor usage and govern access across open and closed models, in public preview starting today.
- The company says AI coding spend has become one of the fastest-growing line items for enterprises, and that developers are left to decide on their own which model fits each task while platform teams juggle providers, budgets and policies.
- Grid's intelligent routing uses Baseten Router, which picks from a team's approved models based on expected performance and cost, including retries, failed tool calls and incomplete answers.
- In initial benchmarks, Baseten says the router achieved nearly 70% cost savings with only a 0.9-percentage-point quality tradeoff versus Opus 5.5, though exact savings vary by workload and desired tradeoffs.
- The gateway covers models from
OpenAI,
Anthropic, Gemini Enterprise Agent Platform,
Amazon Bedrock and Baseten, with team-level access, budgets and company-managed credentials, and developers connect harnesses such as Claude Code, Codex, OpenCode and Pi in a single command.
- Guardrails apply configurable safety policies centrally at the gateway, and Baseten says it is working with Goodfire on Project Beacon to use signals from inside the model to catch unsafe behavior before output reaches a user or tool.





