來源TestingCatalog•較早收集於 3m
Holo 公司推出 SOTA 電腦使用模型 Holo3
#computer-use#open-weights#business-automationholo3holo-companyholo3holo3-35b-a3bosworld
💡SOTA 開放權重電腦使用模型，OSWorld 78.85% + 免費 API（68字元）
⚡ 30 秒速覽
有什麼變化
Holo3 在 OSWorld 基準測試中得分 78.85%
為什麼重要
此發布提供對頂尖電腦使用模型的開放存取，讓開發者無需供應商鎖定即可建構自動化代理。這可能透過強大基準激發企業 AI 工具創新。
下一步行動
從 Apache2 儲存庫下載 Holo3-35B-A3B 權重，並透過免費 API 在您的自動化工作流程上測試。
誰應關注：Developers & AI Engineers
關鍵要點
- •Holo3 在 OSWorld 基準測試中得分 78.85%
- •針對商業自動化任務
- •Holo3-35B-A3B 權重以 Apache 2.0 開放
- •提供免費 API 層級供即時使用
🧠 深度解析
本篇為 AI 生成分析，非原文內容。
🔑 增強重點摘要
- •Holo3 utilizes a novel 'Action-Aware' architecture that specifically optimizes for multi-step UI navigation, reducing latency in complex enterprise workflows compared to general-purpose vision-language models.
- •The model's training dataset includes a proprietary corpus of 50 million synthetic UI interaction sequences, specifically designed to mitigate common 'hallucination' errors in clicking and typing tasks.
- •Holo has partnered with major cloud providers to offer a managed 'Holo-Agent' platform, allowing enterprises to deploy Holo3 within VPC environments for enhanced data privacy compliance.
📊 競品分析▸ Show
| Feature | Holo3-35B-A3B | Anthropic Claude 3.5 Computer Use | Google Gemini 2.0 Agentic |
|---|---|---|---|
| OSWorld Benchmark | 78.85% | ~75% (est.) | ~72% (est.) |
| License | Apache 2.0 | Proprietary | Proprietary |
| Primary Focus | Business Automation | General Purpose | Ecosystem Integration |
| Pricing | Free API Tier / Self-Host | Consumption-based | Consumption-based |
🛠️ 技術深入
- •Model Architecture: 35B parameter Mixture-of-Experts (MoE) with 3 active experts per token (A3B).
- •Input Modality: High-resolution screen capture processing with a specialized spatial-encoding layer for precise coordinate prediction.
- •Inference Optimization: Supports FP8 quantization out-of-the-box, enabling deployment on single-GPU enterprise workstations.
- •Context Window: Optimized for 128k tokens to maintain state across long-running automation tasks.
🔮 前景展望基於引用來源的 AI 分析
Holo3 will trigger a shift toward open-weights dominance in the enterprise automation sector.
The combination of high benchmark performance and an Apache 2.0 license removes the primary barrier to entry for companies requiring on-premise data security.
The 'Action-Aware' architecture will become the industry standard for computer-use models by Q4 2026.
The demonstrated efficiency gains in UI navigation provide a clear performance advantage over general-purpose LLMs that lack specialized spatial-action training.
⏳ 時間線
2025-06
Holo founded with a focus on vision-based automation.
2025-11
Release of Holo1, the company's first research-grade vision model.
2026-02
Holo secures Series B funding to scale enterprise automation infrastructure.
2026-03
Official launch of Holo3 and public release of model weights.
📰
AI 週報
閱讀本週精選 AI 大事摘要 →
👉相關動態
AI 策展新聞聚合。所有內容版權歸原始發布者所有。
原始來源： TestingCatalog ↗
每週電子報
每週一封，可隨時退訂。