H Company

Holo4 27B

H Company

Holo4 27B is H Company’s computer-use vision-language model built on Qwen3.8-27B. It is post-trained to work across graphical interfaces, code, MCP tools, and APIs, using screenshots and tool results to choose its next actions. Released weights support local agent research; an external harness executes clicks, typing, and code and returns observations. The dense 27B model uses a non-commercial weight license, unlike the Apache-licensed sparse Holo4 tier. Its card specifies a 262,144-token configuration limit; this is not a promise of consistent task performance throughout that window.

Holo4 35B-A3B

H Company

Holo4 35B-A3B is H Company’s computer-use vision-language model built on Qwen3.6-35B-A3B. It is post-trained to work across graphical interfaces, code, MCP tools, and APIs, using screenshots and tool results to choose its next actions. Released weights support local agent research; an external harness executes clicks, typing, and code and returns observations. The sparse model activates about 3B language parameters per token but still needs memory for its full checkpoint. Its card specifies a 262,144-token context. Apache-licensed weights and hosted H Models API access are available.

Holotron4 Nano

H Company

Holotron4 Nano is H Company’s computer-use vision-language model built on NVIDIA Nemotron 3 Nano Omni. It is post-trained to work across graphical interfaces, code, MCP tools, and APIs, using screenshots and tool results to choose its next actions. Released weights support local agent research; an external harness executes clicks, typing, and code and returns observations. The sparse model activates about 3B language parameters per token but still needs memory for its full checkpoint. Its card specifies a 262,144-token context. The NVIDIA Open Model Agreement governs its weights; the released page establishes text and image understanding without assuming every modality of its Omni backbone.

Holo3.1

H Company

Holo3.1 is H Company’s updated computer-use model family, adding mobile robustness, native function-calling protocols, and practical local deployment. It spans dense 0.8B, 4B, and 9B Qwen3.5-derived tiers plus a sparse 35B-A3B tier whose current card identifies Qwen3.6 ancestry. The release includes Apache-licensed weights and optimized FP8, NVFP4, and Q4 GGUF options for the larger tier. These are variants of the same trained generation rather than extra catalog models. The current H Models API exposes the 35B tier with a 65,536-token hosted context; that limit is described without treating it as a verified uniform specification for every size.

Holotron 3 Nano

H Company

Holotron 3 Nano is H Company’s Nemotron 3 Nano Omni-derived vision-language policy for computer-use agents. The hybrid Mamba/attention mixture-of-experts backbone has a 30B language tier with about 3B active per token and a native 262,144-token configuration. Post-training on proprietary agent data improves multi-step visual workflows and interface grounding, while the vision encoder supports varied screenshot resolutions. The April 28 release provides downloadable weights under NVIDIA model terms and made the model available in HoloTab. This record covers text and image understanding explicitly described for the released agent model, without assuming every audio/video capability of its Omni backbone.

Holo3 122B-A10B

H Company

Holo3 122B-A10B is H Company’s larger hosted computer-use model, positioned for longer and more complex enterprise workflows. Its mixture-of-experts architecture has 122B total language parameters with about 10B active per token, and combines screenshot understanding with multi-step navigation and structured actions. It was publicly introduced March 31, before the April 1 Hugging Face article. The current provider catalog documents a 65,536-token context, paid API access, and research-only non-commercial weight terms. An openly downloadable 122B checkpoint was not verified, so public access is recorded through the API rather than extrapolated from the Apache-licensed 35B tier.

Holo3 35B-A3B

H Company

Holo3 35B-A3B is H Company’s Apache-licensed sparse vision-language model for enterprise computer use. Based on Qwen3.5-35B-A3B, it is trained on open, synthetic, and human-annotated trajectories to improve perception, multi-step decisions, and navigation across web and desktop software. It was introduced on March 31 alongside a larger hosted Holo3 tier; the subsequent Hugging Face team article appeared April 1. This checkpoint provides downloadable weights and was originally served through the free API tier. Current API documentation instead lists Holo3.1 for the smaller tier, so continued first-party hosted inference on the original Holo3 35B is not asserted.

Holotron-12B

H Company

Holotron-12B is H Company’s first Holotron computer-use model, developed with NVIDIA and post-trained from Nemotron Nano 2 VL. The approximately 12B hybrid state-space/attention vision-language model is optimized for production throughput, long interactive histories, and multiple screenshots. Targeted training turns the general multimodal backbone into an agent policy that interprets screens and emits actions across web, desktop, and mobile environments. The released model card explicitly dates publication to March 16 and identifies NVIDIA model licensing. Public weights allow local experimentation; an external harness supplies screenshots, executes actions, and manages the agent loop. Benchmark throughput is specific to the tested GPU and workload configuration.

Holo2 235B-A22B Preview

H Company

Holo2 235B-A22B Preview is H Company’s separately published large research tier for visual interface localization. Built on Qwen3-VL-235B-A22B-Thinking, it uses sparse language computation with about 22B active parameters per token. The model interprets a textual request and high-resolution screenshot to identify the relevant interface element, and can refine localization through repeated agent steps. The February 3 launch confirms downloadable weights; repository creation in January is not used as publication. It is a language-capable vision model rather than an embedding or detection-only endpoint, but its published evaluation scope centers on grounding instead of broad autonomous-agent performance.

Holo2 30B-A3B

H Company

Holo2 30B-A3B is H Company’s sparse Qwen3-VL-derived model combining visual grounding, navigation, and reasoning for cross-platform computer-use agents. It has a 30B language tier with approximately 3B active parameters per token and is trained on curated localization and agent data. Released weights can drive web, Ubuntu desktop, and Android workflows through an external harness. Unlike the smaller dense Holo2 tiers, its model-card metadata and official launch identify a non-commercial license. This record follows CC-BY-NC-4.0 metadata rather than the card’s conflicting generic Apache text. Sparse computation does not reduce storage to the active parameter count.

Holo2 4B/8B

H Company

The 4B and 8B Holo2 tiers are H Company’s compact, Apache-licensed vision-language models for computer use across web, desktop, and mobile environments. Derived from Qwen3-VL Thinking checkpoints, they combine interface grounding with navigation and self-generated reasoning, improving on earlier Holo localizers that relied on an external navigation model. Training uses curated localization and agent datasets, and the models can operate in a ReAct-style external harness such as Surfer 2. The two dense sizes are grouped here; the larger sparse Holo2 tiers have different licensing and separate records. Public weights support private deployment with compatible inference frameworks.

Holo1.5

H Company

Holo1.5 is H Company’s improved Qwen2.5-VL-based family for interface grounding, UI understanding, and screen question answering, spanning 3B, 7B, and 72B language tiers. It specializes in identifying requested interface elements across browser, desktop, and mobile screens, and can be combined with external navigation models. The provider’s Holo2 announcement explicitly dates its release to September 25, preceding the currently displayed October 9 announcement page. Weights are public, but licenses vary: H Product Research terms for 3B, Apache 2.0 for 7B, and CC-BY-NC-4.0 for 72B. This page groups routine sizes while preserving their different usage terms.

Holo1

H Company

Holo1 is H Company’s original open-weight action vision-language family, with 3B and 7B language-model tiers derived from Qwen2.5-VL. It is trained for web-interface understanding, precise element localization, navigation policy generation, and validating agent progress. The June 3 release includes public weights and a technical paper, earlier than a later provider retrospective’s June 25 date. Holo1 can serve different roles inside the modular Surfer-H harness; the model identity is distinct from that application. Licensing differs by size: the 7B checkpoint uses Apache 2.0, while the 3B repository includes H Product Research terms incorporating the Qwen research license.