开源官方动态Hugging Face Blog2026年9月28日7 分钟阅读

Holo4：为多面手计算机使用代理提供支持

Hugging Face Blog 发布了这条动态，详情请查看官方发布。

Holo4: powering generalist computer-use agents：长流程 Agent 时间线、任务节点与时钟 — AGENT / TIMELINE#25

发布来源

来自 Hugging Face Blog 的官方动态，版本与适用范围以发布方说明为准。

Hugging Face Blog · 查看官方发布

中文机器翻译 · 原文附后

正文 · 来源原文

Holo4是我们新推出的代理模型系列。它有两种尺寸：27B 密集型和 35B-A3B 专家混合型。两者均可在 H 模型 API 上使用。我们还发布了 Holotron 3 的更新版本：Holotron4 Nano。

Holo4 基于我们之前的模型构建，并通过任何可用的接口与软件交互：GUI、代码、MCP 和 API。它在学术基准上得分很高，但我们是为了实际的业务工作流程而构建它的。它是通过对大量环境和任务（包括由我们的代理任务工厂生成的环境和任务）进行监督和强化学习来进行训练的。

立即开始：

🤖 型号：Holo4-27B | Holo4-35B-A3B | Holotron4 Nano 🗂️ 完整合集（FP16、FP8、GGUF）：Holo4 🎞️ 轨迹：查看器 |数据集 ⚡ H 模型 API：快速入门 📝 完整博客文章：hcompany.ai/newsroom/holo4

Holo4 在屏幕上点击和输入，编写并运行自己的代码，并调用 MCP 或 API 工具。它使用适合任务的任何一个。大多数代理模型仅针对一种界面进行训练：以 GUI 为中心的模型在没有屏幕的情况下是盲目的，而更喜欢工具调用的模型则被困在没有 API 的应用程序前面。实际工作并不是这样孤立的，单个业务任务可能需要结合这些不同的方法。

Holo4 可在桌面、网络、Android、代码沙箱和业务 API 上运行。每种情况下都是相同的模型，并且调用方式也相同。您无需为每个平台选择不同的型号。

Holo4 模型比 Qwen 基础有了显着改进。 Holo4 仅在长工作流程上落后于最强的封闭模型：在 OSWorld 2.0 上，Holo4 27B 得分为 61.7%，Opus 5.5 得分为 81.8%，Holo4 35B-A3B 达到 30.9%。然而，它用更少的参数和更低的成本来实现这一点。我们在公共基准测试中开源了分数背后的每条轨迹：在 trajectories.hcompany.ai 上重播每一步或从 Hugging Face 下载它们。

在桌面控制 (OSWorld 2.0) 和 API 使用 (AutomationBench) 的最严格的学术基准测试中，Holo4 以低得多的每项任务成本与前沿模型进行竞争。

性价比图表注释

操作系统世界2.0。成本是根据每次代理运行的输入和输出令牌来估计的。 Holo4 按 H 型号 API 费率定价（单次运行）。 Qwen3.8 27B：模型卡得分，来自我们以阿里云标价运行的代币成本。 Qwen3.6 35B-A3B：在我们的工具中单次运行，按照阿里云标价，缓存命中率为输入价格的 20%。 OpenAI 发布数据提供了 GPT 和 Opus 工作扫描；其他封闭式和开放式重量点使用官方 OSWorld 2.0 排行榜。版本、工具和任务子集有所不同。该线连接封闭模型中的非支配分数和成本对； Holo4 除外。

自动化工作台。 Holo4、Qwen3.8 27B 和 Qwen3.6 35B-A3B：AutomationBench v1.0.6，在我们的内部线束中测量的分数和成本。其他模型：来自 AutomationBench README 的公开集分数、来自官方排行榜的每项任务成本（在私有集上运行）。一旦评估完毕，我们将在私人套件上报告 Holo4。

查看英文原文

Holo4: powering generalist computer-use agents

Hugging Face Blog 发布了这条动态，详情请查看官方发布。

正文 · 来源原文

Holo4 is our new series of agentic models. It comes in two sizes: 27B dense and 35B-A3B Mixture of Experts. Both are available on the H Models API. We are also releasing an updated version of Holotron 3: Holotron4 Nano.

Holo4 builds on our previous model and interacts with software through any available interface: GUIs, code, MCP and APIs. It scores well on academic benchmarks, but we built it for real business workflows. It was trained through supervised and reinforcement learning on a large set of environments and tasks, including those generated by our Agentic Task Factory.

Get started now:

🤖 Models: Holo4-27B | Holo4-35B-A3B | Holotron4 Nano 🗂️ Full collection (FP16, FP8, GGUF): Holo4 🎞️ Trajectories: viewer | dataset ⚡ H Models API: quickstart 📝 Full blog post: hcompany.ai/newsroom/holo4

Holo4 clicks and types on a screen, writes and runs its own code, and calls MCP or API tools. It uses whichever fits the task. Most agentic models are trained for one interface only: GUI-focused models are blind without a screen, while models that prefer tool calling are stuck in front of an application that has no API. Real work is not siloed that way, and a single business task can require combining these different approaches.

Holo4 runs on desktops, on the web, on Android, in a code sandbox and against business APIs. It is the same model in each case and it is called the same way. You do not need to select a different model for each platform.

Notes on the cost-performance charts

OSWorld 2.0. Costs are estimated from the input and output tokens of each agentic run. Holo4 is priced at H Models API rates (single run). Qwen3.8 27B: model card score, cost from the tokens of our run at Alibaba Cloud list prices. Qwen3.6 35B-A3B: single run in our harness, at Alibaba Cloud list prices with cache hits at 20% of the input price. OpenAI launch data supplies the GPT and Opus effort sweeps; other closed and open-weight points use the official OSWorld 2.0 leaderboard. Releases, harnesses and task subsets differ. The line connects non-dominated score and cost pairs among the closed models; Holo4 is excluded.

AutomationBench. Holo4, Qwen3.8 27B and Qwen3.6 35B-A3B: AutomationBench v1.0.6, scores and costs measured in our internal harness. Other models: public-set scores from the AutomationBench README, cost per task from the official leaderboard, which runs on the private set. We will report Holo4 on the private set once it is evaluated.

Trained on environments and tasks from our Agentic Task Factory, Holo4 models excel on professional software. The examples below show Holo4 27B alongside Qwen3.8 27B, its base model. Same prompt and harness for both models.

Build a 3D model of the Eiffel Tower in FreeCAD, at a scale of 1 mm to 1 metre, centred on the origin and aligned to the X and Y axes. Work to this design. The tower is square in plan at every height, never round. Its half-width, measured from the central axis out to the corner, is 62.5 mm at ground level, 32.5 mm at height 57, 17.5 mm at height 115, and 9.35 mm at height 276. Between those heights the half-width follows a smooth curve that falls steeply near the ground and gently higher up, never a straight line. Four identical legs, one per quadrant, each a square column whose outer corner follows that profile. Each leg is 14 mm across at the ground and tapers to 4 mm at height 276. The legs stand apart from the ground up to the first platform, and converge as they rise. Nothing fills the space between them: the tower is open, and you can see straight through it from every side. Three platforms, each a solid square slab centred on the axis: 72 mm across and 4 mm thick at height 57; 40 mm across and 3 mm thick at height 115; 22 mm across and 3 mm thick at height 276. A mast from height 276 to 324, square, 8 mm across at its base tapering to 2 mm at the tip. Every part must be a closed solid with non-zero volume, and no part may fill the space between the legs.

Holo4 27B (84 calls, 1.3M tokens)

Qwen3.8 27B (60 calls, 1.0M tokens)

Build a 3D model in FreeCAD of the H company logo: a solid filled disc beside a blocky sans-serif capital letter H, both extruded to the same thickness, the two shapes of similar height and set apart so they do not overlap, with the centre of the disc level with the middle of the H. Colour both shapes black.

Holo4 27B (94 calls, 1.5M tokens)

Qwen3.8 27B (118 calls, 1.9M tokens)

Build a Pac-Man-style game in Godot and leave it running. A rectangular maze of walls laid out on a grid, with pellets filling every open corridor. A player marker moves continuously along the corridors, eating each pellet it passes over and scoring a point for it. Three ghosts move through the same corridors and chase the player. If a ghost catches the player, the player loses a life and everything resets to its starting position. Score and lives are drawn on screen. No one is going to play this. The player drives itself with a simple heuristic: at each junction it heads toward the nearest pellet, unless a ghost is close, in which case it moves away from the ghost. The game must run unattended and indefinitely, with no keyboard input at all. When it works, start the game and leave it playing.

Holo4 27B (68 calls, 2.4M tokens, 268 lines)

Qwen3.8 27B (197 calls, 11.4M tokens, 327 lines)

Our internal set of agentic pipelines builds interactive environments and verifiable tasks from documentation alone, such as screenshots of real websites or open-source software. So far it has produced about 10,000 tasks across web apps, MCP servers and desktop environments, including hybrid environments that expose the same state through a GUI and MCP.

Alongside training, we rebuilt our harness, the loop that executes the model's actions and manages its context over hundreds of steps, using feedback from agentic performance on OSWorld 2.0. Agents tagged why each task failed and engineers reviewed their fixes. The largest changes were giving the agent a reliable memory that can keep track of hundreds of steps, and a shell on the desktop machine itself.

Opus 5 (70.2%) and GPT-5.6 Sol (66.2%) use max-effort partial rewards on the v2026.08.08 offline set from OpenAI's launch chart, as in the cost-performance plot. Other reference scores come from model cards and the official leaderboard. Task releases, subsets and harnesses vary.

Our post-training stack is designed to adapt to new foundation models and produce agents that generalize across interfaces and environments. As a member of the NVIDIA Nemotron Coalition, we applied our latest stack to the Nemotron 3 Nano Omni model as a follow-up to Holotron 3.

The same recipe turns Nemotron 3 Nano Omni into Holotron4 Nano, a generalist agentic model that significantly improves over the base model on GUI workflows and in environments exposing MCP, APIs or coding sandboxes.

Gains are absolute percentage-point improvements over Nemotron 3 Nano Omni.

These gains show that our recipe transfers well and can turn a generalist model into an agentic expert. Nothing in it is size-specific.

Both sizes are available today on the H Models API. Weights are on Hugging Face in BF16, FP8, NVFP4 and 4-bit GGUF, next to our small model, Holotron4 Nano.

We will release optimized DSpark drafter checkpoints in the coming days to further accelerate inference.

中转查

正文 · 来源原文

性价比图表注释

Holo4: powering generalist computer-use agents

正文 · 来源原文

Notes on the cost-performance charts