{"version":1,"type":"story","url":"https://digestai.news/story/agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool","json":"https://digestai.news/story/agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool.json","markdown":"https://digestai.news/story/agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool.md","slug":"agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool","headline":"Agnes AI releases Agnes-3.0-Flash, a 33B open‑weights multimodal model with tool calling","summary":"Agnes AI announced the open‑weights release of Agnes-3.0-Flash, a 33‑billion‑parameter multimodal model that handles text, image and video inputs while offering built‑in tool‑calling and adjustable reasoning effort. The model uses a hybrid‑attention decoder where three out of four layers run a gated‑delta recurrent rule, leaving only 18 of the 72 layers with a KV cache, which reduces memory pressure and enables a 144‑token context window on modest hardware. \n\nThe team provided a full Hugging Face‑compatible pipeline, including a custom chat template that lets developers select high, medium or low reasoning levels, toggle thinking off, and embed tool definitions that the model emits as structured calls. Sample Docker commands show how to serve the model with sglang, and the release is licensed under Apache 2.0 with a citation pointing to https://agnes‑ai.com/. While the benchmark table is a reference compilation rather than a controlled head‑to‑head test, the authors claim competitive results on reasoning, coding and instruction‑following tasks, positioning Agnes‑3.0‑Flash as a cost‑effective alternative to larger flagship models.","keyPoints":["Agnes-3.0-Flash is a 33B multimodal model supporting text, image, video, and tool‑calling.","Hybrid‑attention decoder reduces KV‑cache layers to 18, enabling a 144‑token context window on cheaper hardware.","Released under Apache 2.0 with Hugging Face integration and Docker serving scripts."],"whyItMatters":"A high‑performing, open‑weights multimodal model lowers the barrier for developers to run advanced AI reasoning without expensive GPU clusters, accelerating research and product prototyping.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["Agnes AI","LMSYS"],"models":["Agnes-3.0-Flash"],"people":[]},"firstPublishedAt":"2026-09-12T08:05:53Z","updatedAt":"2026-09-12T08:05:53Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"huggingface.co","title":"Agnes-AI/Agnes-3.0-Flash 33B Multimodal, AA score: 36","url":"https://huggingface.co/Agnes-AI/Agnes-3.0-Flash","publishedAt":"2026-09-12T08:05:53Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[{"site":"Reddit","url":"https://www.reddit.com/r/LocalLLaMA/comments/1we6lrn/agnesaiagnes30flash_33b_multimodal_aa_score_36/","points":null}],"thread":null,"cite":{"text":"Digest AI, \"Agnes AI releases Agnes-3.0-Flash, a 33B open‑weights multimodal model with tool calling\", 12 September 2026, https://digestai.news/story/agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool","publisher":"Digest AI","title":"Agnes AI releases Agnes-3.0-Flash, a 33B open‑weights multimodal model with tool calling","datePublished":"2026-09-12T08:05:53Z","url":"https://digestai.news/story/agnes-ai-releases-agnes-3-0-flash-a-33b-openweights-multimodal-model-with-tool"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}