Case Study Deskcasestudydesk.com
MediaSourced

Case Study: The Washington Post runs Ask The Post AI on Together AI at 2-second responses

The Washington Post Case StudySourced & dated by Case Study Desk
Key facts · TL;DR
Company
The Washington Post
Industry
Media
Challenge
Journalism AI needed reliable, cost-predictable inference with model control
Headline result
Reliable open-model inference for a public-facing journalism assistant

Key results

2 sec
model response time
Ask The Post AI
1.79B
monthly input tokens processed

The challenge

The Washington Post launched Ask The Post AI, a public-facing assistant answering reader questions from its journalism. It required consistent response times and predictable costs without depending on proprietary model APIs.

The solution

The Post ran open-source Llama and Mistral models on Together AI with customized configurations for its journalism use case and fixed monthly pricing, retaining full model ownership.

Together AI's flexibility in creating customized configurations for our journalism use case has been invaluable to our success.

SH
Sam Han
Chief AI Officer, The Washington Post

The results, in context

Together AI's page reports the Post achieved consistent roughly 2-second responses while processing 1.79 billion input tokens per month, maintaining enterprise-grade reliability for the public assistant.

Products used

Together AI Dedicated EndpointsTogether AI Serverless Inference