Baseten and Postman: Building reliable, cost-efficient agents for the enterprise

The Agent Stack

If you do not control your agents, your agents control your company.

Postman’s Agent Stack helps enterprises prepare APIs for agents, control which systems agents can access, and govern their actions. It includes four products.

Passport: Secures API access for humans and agents using scoped credential references, without exposing the underlying API keys.

Fabric: Acts as an AI-native gateway that routes and governs AI, MCP, and API traffic.

Astropods: Serves as an operating system with runtime, observability, and guardrails for your AI agents in production.

Fern: Generates documentation, SDKs, and CLIs that make APIs discoverable, typed, and callable by developers and AI agents.

While Postman gives you the tools to build and operate agents, Baseten provides the performant, cost-efficient infrastructure they run on.

Fast, reliable, cost-efficient inference

Postman provides tools to build agents and connect them to APIs. As Postman’s recommended inference partner for enterprise teams, Baseten runs the models behind those agents with fast inference, resilient multi-cloud capacity, and the post-training infrastructure needed to specialize models.

Fast: Model APIs deliver high throughput and low latency, with seven 2026 launches ranking among the top two providers on Artificial Analysis for both throughput and latency.⁠

• Beam 1.0, #1 in throughput and latency

• Nemotron 3.5 Super VL, #2 in throughput, #1 in latency

• Muse Spark Open, #2 in throughput and latency

• DeepSeek V4.1 Flash 1017, #2 in throughput and latency

• DeepSeek V4.1 Flash, #1 in throughput and latency

• DeepSeek V4 Flash, #1 in throughput and latency

• GLM-5.2 Fast, #2 in throughput and latency

Reliable: Multi-cloud Capacity Management (MCM) pools GPU capacity across clouds, unlocking massive capacity to serve enterprise workloads, with built-in resilience to hardware failures. Baseten has zero data retention (ZDR) by default, with the ability to restrict or customize the clouds, regions, and clusters you run on, including single-tenant and active-active deployments.⁠

Cost-efficient: Baseten Training provides infrastructure and research support to post-train open models for your agent’s specific tasks, matching or exceeding the quality of large frontier models at a fraction of the cost.

• Harvey post-trained a 27B open model to match the closed-source frontier in performance on its legal agent benchmark, where top closed models cost roughly $50 per task.

Post-train models for better response quality

For the majority of economically valuable tasks, the gap between what a general-purpose model can do via prompting and what a specialized model can do via post-training is large.

Post-training also improves your agent’s response quality, while enabling you to use a faster, cheaper model. Here’s how it works with Baseten:

• Train for specific tasks. Adapt models to the agent’s workflow with supervised fine-tuning and reinforcement learning.

• Deploy and iterate. Deploy trained models on Baseten and leverage usage data to guide further training.

• Work with experts. Receive direct support from Baseten’s post-training team.

• Keep ownership. Retain your model weights, evaluations, and training scripts.

Leverage cost-efficient agents with Baseten and Postman

Baseten gives us the infrastructure and post-training expertise to put the right model on every job in Agent Mode, with the economics to put more AI into the hands of enterprises.

Building with Postman and preparing to deploy to production? Connect with us to explore how Baseten inference and post-training can help you optimize reliability, cost efficiency, and performance across your agent workloads.

Build with Postman. Run on Baseten.