--- title: YOLO26 Open-Vocabulary Object Detection with YOLOE-26 description: YOLOE-26 extends the open-vocabulary detection capability of YOLOE to the newer YOLO26 architecture, letting developers detect object classes beyond a fixed... canonical: https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: YOLO26 Open-Vocabulary Object Detection with YOLOE-26 | daily.dev og:description: YOLOE-26 extends the open-vocabulary detection capability of YOLOE to the newer YOLO26 architecture, letting developers detect object classes beyond a fixed... og:url: https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx og:image: https://api.daily.dev/og/posts/Rpj3KFESX.png og:image:alt: YOLO26 Open-Vocabulary Object Detection with YOLOE-26 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. # YOLO26 Open-Vocabulary Object Detection with YOLOE-26 **[PyImageSearch](https://daily.dev/sources/pyimagesearch)** · 31 min read · 1 upvotes · 0 comments ## Summary YOLOE-26 extends the open-vocabulary detection capability of YOLOE to the newer YOLO26 architecture, letting developers detect object classes beyond a fixed closed-set label list using text prompts, visual (reference box) prompts, or prompt-free inference against a built-in vocabulary. The lesson walks through the RepRTA, SAVPE, and LRPC modules that power each mode, shows Ultralytics code examples for each prompting style, covers benchmark numbers (up to 40.6 AP on LVIS with text prompts), discusses when to choose YOLOE-26 versus a standard closed-set YOLO26 model, and lists common failure modes and fixes such as overly broad or obscure prompts, poor reference boxes, and semantic overlap between prompts. It also notes a deployment detail: exported YOLOE models freeze their configured prompt classes, requiring re-export to change them. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the difference between YOLOE-26 text prompting and visual prompting in Ultralytics? Text prompting uses set_classes() with word labels and the RepRTA module to align text embeddings with visual region features, while visual prompting uses SAVPE with reference bounding boxes (via YOLOEVPSegPredictor) so the model finds regions visually similar to a shown example. Text prompting suits concepts that are easy to name; visual prompting suits objects that are easier to show than describe, such as specific industrial parts. _daily.dev surfaces practical comparisons like this for developers choosing between prompting strategies in vision projects._ ### How much accuracy do you lose using YOLOE-26 prompt-free mode compared to text or visual prompting? On the LVIS minival benchmark, the x-scale YOLOE-26 model reaches 40.6 AP with text prompts, 38.5 AP with visual prompts, but only 31.1 AP in the prompt-free non-E2E setting. Prompt-free mode trades that accuracy for convenience since it relies on a built-in vocabulary via LRPC rather than an explicit prompt at inference time. _Developers weighing convenience against accuracy in detection pipelines can track benchmark tradeoffs like this on daily.dev._ ### Can you change the class prompts of a YOLOE model after exporting it? No, once a YOLOE model is exported, the configured prompt classes are baked into the exported weights and cannot be swapped afterward. To change the detected classes, you must re-export from the original checkpoint rather than modify the already-exported artifact, which matters when moving from experimentation to a deployable production model. _Teams planning deployment pipelines around promptable detectors can keep such export gotchas in view via daily.dev._ ## Similar posts on daily.dev - [Train YOLO26 on a Custom Dataset with YOLOE-26 Auto-Labeling](https://daily.dev/posts/train-yolo26-on-a-custom-dataset-with-yoloe-26-auto-labeling-gyjcey2di) · PyImageSearch · 0 upvotes · 0 comments - [Train YOLO26 for Retail Object Detection on DigitalOcean GPUs](https://daily.dev/posts/train-yolo26-for-retail-object-detection-on-digitalocean-gpus-8e9osvjze) · DigitalOcean Community · 0 upvotes · 0 comments - [Fine-tune Ultralytics YOLO26 Object Detection Model](https://daily.dev/posts/fine-tune-ultralytics-yolo26-object-detection-model-ktflc0ns2) · Daily Dose of Data Science \| Avi Chawla \| Substack · 0 upvotes · 0 comments - [Best Object Detection Models for Machine Learning in 2026](https://daily.dev/posts/best-object-detection-models-for-machine-learning-in-2026-jzubdcufk) · JetBrains · 0 upvotes · 0 comments - [YOLOv2 & YOLO9000 Paper Walkthrough: Better, Faster, Stronger](https://daily.dev/posts/yolov2-yolo9000-paper-walkthrough-better-faster-stronger-znlh2qfuv) · Towards Data Science · 2 upvotes · 0 comments --- Tags: [#computer-vision](https://daily.dev/tags/computer-vision), [#object-detection](https://daily.dev/tags/object-detection) [View this post on daily.dev](https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx) ```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":"YOLO26 Open-Vocabulary Object Detection with YOLOE-26","url":"https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx"},"datePublished":"2026-08-24T12:49:39.105Z","dateModified":"2026-09-14T06:55:26.045Z","description":"YOLOE-26 extends the open-vocabulary detection capability of YOLOE to the newer YOLO26 architecture, letting developers detect object classes beyond a fixed...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/674c0657f005f7a48594cded03de0bf5?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/674c0657f005f7a48594cded03de0bf5?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"PyImageSearch","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":"PyImageSearch","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/33ddc48b99ee47eb9283ef1fdcff3095","url":"https://daily.dev/sources/pyimagesearch"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"computer-vision,object-detection","timeRequired":"PT31M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"PyImageSearch","item":"https://daily.dev/sources/pyimagesearch"},{"@type":"ListItem","position":3,"name":"YOLO26 Open-Vocabulary Object Detection with YOLOE-26"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/yolo26-open-vocabulary-object-detection-with-yoloe-26-rpj3kfesx#faq","mainEntity":[{"@type":"Question","name":"What is the difference between YOLOE-26 text prompting and visual prompting in Ultralytics?","acceptedAnswer":{"@type":"Answer","text":"Text prompting uses set_classes() with word labels and the RepRTA module to align text embeddings with visual region features, while visual prompting uses SAVPE with reference bounding boxes (via YOLOEVPSegPredictor) so the model finds regions visually similar to a shown example. Text prompting suits concepts that are easy to name; visual prompting suits objects that are easier to show than describe, such as specific industrial parts. daily.dev surfaces practical comparisons like this for developers choosing between prompting strategies in vision projects."}},{"@type":"Question","name":"How much accuracy do you lose using YOLOE-26 prompt-free mode compared to text or visual prompting?","acceptedAnswer":{"@type":"Answer","text":"On the LVIS minival benchmark, the x-scale YOLOE-26 model reaches 40.6 AP with text prompts, 38.5 AP with visual prompts, but only 31.1 AP in the prompt-free non-E2E setting. Prompt-free mode trades that accuracy for convenience since it relies on a built-in vocabulary via LRPC rather than an explicit prompt at inference time. Developers weighing convenience against accuracy in detection pipelines can track benchmark tradeoffs like this on daily.dev."}},{"@type":"Question","name":"Can you change the class prompts of a YOLOE model after exporting it?","acceptedAnswer":{"@type":"Answer","text":"No, once a YOLOE model is exported, the configured prompt classes are baked into the exported weights and cannot be swapped afterward. To change the detected classes, you must re-export from the original checkpoint rather than modify the already-exported artifact, which matters when moving from experimentation to a deployable production model. Teams planning deployment pipelines around promptable detectors can keep such export gotchas in view via daily.dev."}}]} ```