{"id":46997,"date":"2026-06-23T12:44:06","date_gmt":"2026-06-23T12:44:06","guid":{"rendered":"https:\/\/www.gsma.com\/solutions-and-impact\/gsma-open-gateway\/?p=46997"},"modified":"2026-06-23T12:49:31","modified_gmt":"2026-06-23T12:49:31","slug":"open-telco-ai-gemma","status":"publish","type":"post","link":"https:\/\/www.gsma.com\/solutions-and-impact\/gsma-open-gateway\/open-telco-ai-gemma\/","title":{"rendered":"Accelerating Telecom Innovation: How AT&T and the GSMA are Advancing Open Telco Models with Google\u2019s Gemma"},"content":{"rendered":"\n

The Specialised Challenge of Telecom AI<\/em><\/p>\n\n\n\n

Telecommunications is an incredibly complex, highly specialised domain. Modern mobile networks are inherently multivendor, featuring diverse and often proprietary data structures. While AI has made massive leaps in general language and coding, telecom domain knowledge is rarely accessible on the open internet\u2014there is simply no “Wikipedia” for telecoms.<\/p>\n\n\n\n

This data scarcity creates a major hurdle for AI models trying to deeply understand network operations. Operating at an immense global scale, connecting billions of people hundreds of billions of times a day, the industry requires absolute precision. Yet, according to GSMA Intelligence, only 16% of total AI deployments in telecoms are on the network<\/a>, largely due to the difficulty of training models on specialised domain knowledge.<\/p>\n\n\n\n


Why Domain-Specific Models Matter<\/strong><\/p>\n\n\n\n

While today’s general frontier models are incredibly capable at broad reasoning and language tasks, they lack the foundational context required to manage critical infrastructure. General models struggle with the highly specialised vocabulary, complex network topologies, and vendor-specific telemetry data unique to the telecom sector. Telco-specific models solve this by anchoring the AI in the actual realities of network operations. By training on domain-specific datasets, these tailored models can interpret nuanced technical logs, diagnose network performance bottlenecks, and understand standard industry protocols with the high degree of accuracy and precision required for real-time systems.<\/p>\n\n\n\n

Google\u2019s Gemma Model: Driving the Open Telco AI Initiative<\/strong><\/h2>\n\n\n\n

To address this challenge, the GSMA recently launched the Open Telco AI<\/a> platform to build accurate, efficient, and trusted telco-grade AI. As a core part of this collaborative effort, AT&T post-trained a family of open telco models, called OTel<\/a>, on different architectures including Google\u2019s open source models called Gemma.<\/p>\n\n\n\n

These models were trained on a specialised telco-specific dataset curated by GSMA and its collaborators, including telecom operators, network equipment providers, and academia. The initiative successfully delivered 30 models across a range of sizes and architectures, optimising the balance between accuracy versus efficiency.<\/p>\n\n\n\n

Crucially, these models are built with safety at their core, being trained for abstention using retrieval augmented generation (RAG) to drastically reduce hallucinations\u2014an absolute necessity in highly regulated telecom environments.<\/p>\n\n\n\n

Key Findings: Gemma Emerging as a Leader<\/strong><\/p>\n\n\n\n

The findings highlight the strength of Gemma compared to other architectures, demonstrating strong performance gains across the entire OTel model family after telecom-specific fine-tuning. Notably: <\/p>\n\n\n\n