Doubleword & .txt partner to provide structured generation outputs natively through Doubleword

Doubleword x .txt: Structured outputs from any model, delivered at scale.

We built Doubleword and .txt on the same bet: that the future runs on open-source models, and that how you run them matters as much as which one you pick.

.txt focused on what comes out. We created Outlines, the open-source library that pioneered structured generation, starting with forcing LLMs to produce JSON objects that exactly match a schema. Perfect adherence. Better accuracy.

Doubleword focused on what's underneath: inference infrastructure purpose-built for high-volume workloads, allowing teams to scale without scaling their costs.

The workloads that need structured generation most (data extraction pipelines, agent tool calls, and any workflow where a malformed response causes a process to crash) are also the highest-volume inference workloads. So we picked up the phone.

Today, we're announcing that .txt's structured generation engine is now directly available through Doubleword. Doubleword customers can now benefit from the most correct and complete implementation of structured outputs available, on a pay-per-token basis. No more bad outputs!

[Get started with some free credits →] https://app.doubleword.ai