{"id":146725,"date":"2026-08-20T23:29:09","date_gmt":"2026-08-20T23:29:09","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/146725\/"},"modified":"2026-08-20T23:29:09","modified_gmt":"2026-08-20T23:29:09","slug":"att-slashes-ai-costs-with-model-routers-and-open-source","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/146725\/","title":{"rendered":"AT&#038;T Slashes AI Costs With Model Routers and Open Source"},"content":{"rendered":"<p><a href=\"https:\/\/www.att.com\/\" rel=\"nofollow noopener\" target=\"_blank\">AT&amp;T<\/a> cut the costs of coding and some other advanced artificial intelligence tasks by as much as 56% by using tools that route employees\u2019 queries to cheaper models when appropriate, The Information <a href=\"https:\/\/www.theinformation.com\/newsletters\/applied-ai\/t-using-open-source-models-curb-anthropic-bills\" rel=\"nofollow noopener\" target=\"_blank\">reported<\/a> Thursday (Aug. 20).<\/p>\n<p>When doing so, the quality of the performance of the AI declined by only 2%, according to the report, which cited an interview with <a href=\"https:\/\/www.linkedin.com\/in\/markaustin5?trk=public_post_feed-actor-name\" rel=\"nofollow noopener\" target=\"_blank\">Mark Austin<\/a>, an AT&amp;T vice president who oversees AI used by the telecommunication company\u2019s employees.<\/p>\n<p>The tools AT&amp;T used in this instance were <a href=\"https:\/\/www.litellm.ai\/\" rel=\"nofollow noopener\" target=\"_blank\">LiteLLM<\/a> model routers, which determine the complexity of the task and then decide if it can be sent to a cheaper AI model, the report said.<\/p>\n<p>AT&amp;T aims to keep its employees spending on models from <a href=\"https:\/\/www.anthropic.com\/\" rel=\"nofollow noopener\" target=\"_blank\">Anthropic<\/a> and <a href=\"https:\/\/openai.com\/\" rel=\"nofollow noopener\" target=\"_blank\">OpenAI<\/a> flat by making greater use of open-source or open-weight models. The company aims to increase the share of employees\u2019 queries that are powered by open-source models from the current 40% to between 60% and 70% in the coming years, per the report.<\/p>\n<p>AT&amp;T is using open-source or open-weight models such as <a href=\"https:\/\/www.nvidia.com\/en-us\/\" rel=\"nofollow noopener\" target=\"_blank\">Nvidia<\/a>\u2019s Nemotron, <a href=\"https:\/\/www.meta.com\/about\/?srsltid=AfmBOoogX0pzv0NKlLlzT6ogSSTDT4bJrqicvcQCsOIh4kMEmgmo6sOe\" rel=\"nofollow noopener\" target=\"_blank\">Meta<\/a>\u2019s Llama and <a href=\"https:\/\/about.google\/\" rel=\"nofollow noopener\" target=\"_blank\">Google<\/a>\u2019s Gemma. The company is not using open-source models from the Chinese firms <a href=\"https:\/\/www.deepseek.com\/en\/\" rel=\"nofollow noopener\" target=\"_blank\">DeepSeek<\/a> and <a href=\"https:\/\/www.moonshot.ai\/\" rel=\"nofollow noopener\" target=\"_blank\">Moonshot<\/a> but is evaluating the potential risks of using them, according to the report.<\/p>\n<p>AT&amp;T\u2019s experience has been that the capabilities of open-source models have generally been six to 10 months behind those of frontier models, but the gap is narrowing and the open-source models are \u201cjust as good or better\u201d than older models from Anthropic and OpenAI, Austin said, per the report.<\/p>\n<p>It was reported in June that companies are looking to better manage their <a href=\"https:\/\/www.pymnts.com\/news\/artificial-intelligence\/2026\/soaring-ai-costs-push-enterprise-buyers-to-cheaper-chinese-models\/\" rel=\"nofollow noopener\" target=\"_blank\">use of AI<\/a> after seeing the costs of the technology rise.<\/p>\n<p>The rising costs have been driven by the shift from chatbots to agents, which consume more computing power, as well as the AI labs\u2019 move from flat subscriptions to token-based billing.<\/p>\n<p>PYMNTS reported in July that new tools are being launched to address the issue of <a href=\"https:\/\/www.pymnts.com\/news\/artificial-intelligence\/2026\/finance-teams-are-done-flying-blind-on-ai-costs\/\" rel=\"nofollow noopener\" target=\"_blank\">AI costs<\/a> and that the era of \u201ctokenmaxxing,\u201d or pushing employees toward the biggest AI models and heaviest usage, is ending after two years of unchecked growth.<\/p>\n","protected":false},"excerpt":{"rendered":"AT&amp;T cut the costs of coding and some other advanced artificial intelligence tasks by as much as 56%&hellip;\n","protected":false},"author":2,"featured_media":146726,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[24,25,28811,66,2415,310,311],"class_list":["post-146725","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","tag-ai","tag-artificial-intelligence","tag-att","tag-news","tag-open-source-ai","tag-pymnts-news","tag-whats-hot"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/146725","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=146725"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/146725\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/146726"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=146725"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=146725"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=146725"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}