{"version":1,"type":"story","url":"https://digestai.news/story/h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks","json":"https://digestai.news/story/h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks.json","markdown":"https://digestai.news/story/h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks.md","slug":"h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks","headline":"H company launches Holo4 agentic models for desktop and API tasks","summary":"H company announced the Holo4 series, offering two agentic models—Holo4‑27B (dense) and Holo4‑35B‑A3B (Mixture of Experts)—through its H Models API. An updated Holotron4 Nano is also released. The models can interact with software via GUIs, code, MCP and APIs, aiming at real business workflows rather than single‑interface tasks.\n\nOn the OSWorld 2.0 benchmark Holo4‑27B achieved a 61.7% score, trailing the closed‑source Opus 5.5 at 81.8%, while the larger Holo4‑35B‑A3B scored 30.9%. Despite lower scores, Holo4 delivers comparable performance at a much lower cost per task. The company also applied its stack to NVIDIA’s Nemotron 3 Nano Omni, producing the Holotron4 Nano model that improves on the base model across GUI, API and coding environments.\n\nAll model weights are available on Hugging Face in multiple formats (BF16, FP8, NVFP4, 4‑bit GGUF), and the company plans to release optimized DSpark drafter checkpoints soon, further accelerating inference for developers.","keyPoints":["Holo4 released in two sizes, 27B dense and 35B‑A3B Mixture of Experts, on H Models API.","On OSWorld 2.0 benchmark Holo4‑27B scores 61.7% versus Opus 5.5’s 81.8%, while Holo4‑35B‑A3B scores 30.9%.","Holo4 can click, type, write code and call APIs across desktop, web, Android and sandbox environments using a single model."],"whyItMatters":"By handling GUIs, code and API calls with one model, Holo4 lowers the cost and complexity of building general‑purpose AI agents for business workflows, challenging larger closed models.","category":{"slug":"agents","name":"Agents & Tools","url":"https://digestai.news/category/agents"},"entities":{"companies":["H company","NVIDIA"],"models":["Holo4-27B","Holo4-35B-A3B","Holotron4 Nano","Nemotron 3 Nano Omni"],"people":[]},"firstPublishedAt":"2026-09-28T09:44:05Z","updatedAt":"2026-09-28T09:44:05Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"Hugging Face","title":"Holo4: powering generalist computer-use agents","url":"https://huggingface.co/blog/Hcompany/holo4","publishedAt":"2026-09-28T09:44:05Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"H company launches Holo4 agentic models for desktop and API tasks\", 28 September 2026, https://digestai.news/story/h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks","publisher":"Digest AI","title":"H company launches Holo4 agentic models for desktop and API tasks","datePublished":"2026-09-28T09:44:05Z","url":"https://digestai.news/story/h-company-launches-holo4-agentic-models-for-desktop-and-api-tasks"},"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"}