--- title: Am I paying for a giant model I don't need? | daily.dev description: A walkthrough demonstrates using Microsoft Foundry's model router to route agent prompts across a family of GPT 5.6 models (Luna, Terra, Soul) to cut costs... canonical: https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Am I paying for a giant model I don't need? | daily.dev og:description: A walkthrough demonstrates using Microsoft Foundry's model router to route agent prompts across a family of GPT 5.6 models (Luna, Terra, Soul) to cut costs... og:url: https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj og:image: https://api.daily.dev/og/posts/rOTxw8JLj.png og:image:alt: Am I paying for a giant model I don't need? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Am I paying for a giant model I don't need? **[Microsoft Developer](https://daily.dev/sources/microsoftdeveloper)** · 4 min read · 0 upvotes · 0 comments ## Summary A walkthrough demonstrates using Microsoft Foundry's model router to route agent prompts across a family of GPT 5.6 models (Luna, Terra, Soul) to cut costs instead of always calling the largest model. Using the open source model router auto evaluation tool, the author sets up a baseline and judge model, runs a sample evaluation in VS Code, and reviews a generated dashboard. Results on a 10-prompt sample showed a 37.7% cost savings, slightly slower average latency, and a small dip in quality compared to using GPT 5.6 Soul alone. The author notes 10 prompts is too small a sample for real judgment and recommends running 30+ prompts and weighing quality, latency, and cost based on your agent's priorities. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much cost savings can I expect from using Microsoft Foundry model router instead of a single GPT 5.6 model? In one evaluation run across 10 prompts, routing across the GPT 5.6 family (Luna, Terra, Soul) produced a 37.7% cost savings compared to calling GPT 5.6 Soul directly. However average latency was slightly slower and there was a small dip in quality win rate, so results should be validated with a larger prompt set, ideally 30 or more prompts, before deciding. _Teams weighing model router tradeoffs for their own agents can track real-world cost and quality benchmarks on daily.dev._ ### How do I evaluate whether Microsoft Foundry model router is worth using for my agent? Use the open source model router auto evaluation project, which automates quality, cost, and latency comparisons between model router and a baseline model. Clone the repo, set up a virtual environment, configure endpoints, API keys, deployment names, and pricing entries in default.yaml, then run python scripts/run_eval.py with either the included 10-prompt sample set or a custom dataset, and review the generated dashboard. _daily.dev helps developers comparing agent model strategies stay current on tools like model router auto evaluation._ ## Similar posts on daily.dev - [More models, more choice: Comparing 11 different AI models](https://daily.dev/posts/more-models-more-choice-comparing-11-different-ai-models-uqrz8i7ag) · Hacker News · 1 upvotes · 0 comments - [Foundry Model Router Expands from Two Regions to 28, Refreshing Its Model Pool](https://daily.dev/posts/foundry-model-router-expands-from-two-regions-to-28-refreshing-its-model-pool-ooyd7pizf) · InfoQ · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#gpt](https://daily.dev/tags/gpt) [View this post on daily.dev](https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Am I paying for a giant model I don't need?","url":"https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj"},"datePublished":"2026-10-01T23:25:14.401Z","dateModified":"2026-10-01T23:25:32.216Z","description":"A walkthrough demonstrates using Microsoft Foundry's model router to route agent prompts across a family of GPT 5.6 models (Luna, Terra, Soul) to cut costs...","image":"https://i.ytimg.com/vi/MOMk8rPUB4w/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/MOMk8rPUB4w/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Microsoft Developer","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Microsoft Developer","logo":"https://media.daily.dev/image/upload/s--xZzU3zUq--/f_auto/v1746714872/logos/microsoftdeveloper","url":"https://daily.dev/sources/microsoftdeveloper"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,gpt","timeRequired":"PT4M","video":{"@type":"VideoObject","name":"Am I paying for a giant model I don't need?","description":"A walkthrough demonstrates using Microsoft Foundry's model router to route agent prompts across a family of GPT 5.6 models (Luna, Terra, Soul) to cut costs...","thumbnailUrl":"https://i.ytimg.com/vi/MOMk8rPUB4w/sddefault.jpg","uploadDate":"2026-10-01T23:25:14.401Z","duration":"PT4M","url":"https://api.daily.dev/r/rOTxw8JLj","embedUrl":"https://www.youtube.com/embed/MOMk8rPUB4w"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Microsoft Developer","item":"https://daily.dev/sources/microsoftdeveloper"},{"@type":"ListItem","position":3,"name":"Am I paying for a giant model I don't need?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/am-i-paying-for-a-giant-model-i-don-t-need--rotxw8jlj#faq","mainEntity":[{"@type":"Question","name":"How much cost savings can I expect from using Microsoft Foundry model router instead of a single GPT 5.6 model?","acceptedAnswer":{"@type":"Answer","text":"In one evaluation run across 10 prompts, routing across the GPT 5.6 family (Luna, Terra, Soul) produced a 37.7% cost savings compared to calling GPT 5.6 Soul directly. However average latency was slightly slower and there was a small dip in quality win rate, so results should be validated with a larger prompt set, ideally 30 or more prompts, before deciding. Teams weighing model router tradeoffs for their own agents can track real-world cost and quality benchmarks on daily.dev."}},{"@type":"Question","name":"How do I evaluate whether Microsoft Foundry model router is worth using for my agent?","acceptedAnswer":{"@type":"Answer","text":"Use the open source model router auto evaluation project, which automates quality, cost, and latency comparisons between model router and a baseline model. Clone the repo, set up a virtual environment, configure endpoints, API keys, deployment names, and pricing entries in default.yaml, then run python scripts/run_eval.py with either the included 10-prompt sample set or a custom dataset, and review the generated dashboard. daily.dev helps developers comparing agent model strategies stay current on tools like model router auto evaluation."}}]} ```