Baseten Partners with OpenAI to Bring Open-Model Inference to Codex and Responses API
Summary
- • Baseten joins OpenAI B2B Marketplace as one of the first open-model inference providers
- • Enterprise customers can apply existing OpenAI spend commitments to open models served by Baseten
- • Partnership enables multi-model routing between open and closed models in Codex coding workflows
- • Baseten runs 90+ clusters across 20+ clouds with zero data retention and same-day model access
Details
Baseten joins OpenAI B2B Marketplace as an open-model inference provider
Baseten is one of the first open-model inference providers to join the OpenAI B2B Marketplace, with a native integration within Codex and the Responses API for OpenAI enterprise customers.
Enterprise customers can redirect existing OpenAI spend to open models via Baseten
OpenAI enterprise customers can now apply existing OpenAI spend commitments to open models served by Baseten, allowing cost-optimized multi-model routing without new procurement contracts.
Baseten runs 90+ clusters across 20+ clouds with 200+ compute partners active-active
Baseten's infrastructure operates active-active across 90+ clusters on 20+ clouds with relationships with 200+ compute providers, automatically rerouting traffic on provider or regional failure without service interruption.
Day-zero open model access — new frontier models available to deploy the day they ship
Baseten deploys major open models on its Model APIs the day they are publicly released, so enterprises can evaluate and integrate the latest frontier models immediately with a single code change.
Zero data retention and US-based infrastructure for enterprise compliance requirements
All inference on Baseten runs on US-based infrastructure with zero data retention (ZDR) on all prompts, with options for region pinning and fine-grained authentication and authorization for higher compliance needs.
Blaxel (now part of Baseten) provides code-execution sandboxes for AI coding agents
Blaxel, recently acquired by Baseten, gives every AI coding agent its own sandbox environment for running and testing generated code — a critical capability for agentic coding workflows.
Details of Baseten's partnership with OpenAI and the open-model inference integration for Codex and Responses API.
What This Means
Baseten's partnership with OpenAI creates a practical bridge between enterprise OpenAI commitments and the growing ecosystem of open models, letting companies optimize AI coding workflows without new procurement overhead. For organizations already spending on OpenAI, routing tasks to best-fit open models through the same commitment unlocks meaningful cost savings and performance flexibility. This signals a broader market shift toward multi-model architectures where AI coding — currently the most widely deployed AI use case — is increasingly a mix of proprietary and open-source models. Infrastructure providers like Baseten are emerging as critical middleware in this landscape.