Holotron Models by H Company

H CompanyNVIDIA Open ModelOpen weightsAgentsMultimodal

H Company releases · 12 in the last 12 months · this family litChangelog →

2 models2026Up to 256K ctx

Last refreshed 2026-06-07. Next refresh: weekly.

Details

ResearcherH Company

LicenseNVIDIA Open Model

Commercial usePermitted

Models2

Released2026

Max context256K

Capabilities

VisionAll models

MultimodalAll models

Links

Website HuggingFace

About

H Company's Holotron model series, built on NVIDIA Nemotron architecture with Mamba-2 SSM layers for high-throughput computer-use agents. Features native long-context support (256K tokens) and the NVIDIA Open Model License.

Decision facts

Best fit: agentscomputer usevision and multimodal work
Capability starting point: Holotron-3-Nano with 256K context and multimodal inputs
Lowest tracked input: Not tracked
Closest related family: Agents-A1

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

2 in view

Holotron-3-NanoCurrent

Use when the workload needs computer use, 256K context, and 30 parameters.

2026-04computer use256K context30 parameters

Holotron-12BCurrent

Use when the workload needs computer use, 128K context, and 12 parameters.

2026-03computer use128K context12 parameters

Current Holotron variants with use-when guidance and lifecycle status
Model	Use when	Released	Signals	Status
Holotron-3-Nano	Use when the workload needs computer use, 256K context, and 30 parameters.	2026-04	computer use256K context30 parameters	Current
Holotron-12B	Use when the workload needs computer use, 128K context, and 12 parameters.	2026-03	computer use128K context12 parameters	Current