AT&T routes 40% of AI requests to open-source models, cutting costs

AT&T is significantly shifting its AI strategy by routing 40% of employee AI requests through open-source models, with plans to increase this to 60-70%. This move has reduced costs for coding and advanced tasks by up to 56%, with only a minor 2% decrease in output quality. The company is utilizing models from Nvidia, Google's Gemma, and Meta's Llama, and has also fine-tuned a telecom-specific model called OTel 2.0 based on Gemma 4 31B. AI

IMPACT Demonstrates a viable strategy for enterprises to reduce AI operational costs by leveraging open-source models.

RANK_REASON Company's adoption of open-source models for internal AI requests.

Read on Mastodon — fosstodon.org →

• AT&T
• Gemma
• Llama
• Mastodon
• Nvidia

• product
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →