The Open Weights
LatestModelsLeaderboardsCompanies
Subscribe
The Open Weights

The daily record of open-source AI. New model releases, leaderboards, and what's coming next — written for people who ship.

Refreshed every 12 hours

Discover

  • Latest releases
  • New today
  • Trending models

Browse

  • All models
  • Companies
  • Categories
  • Leaderboards

About

  • About
  • Editorial policy
  • Hardware estimates
  • RSS feed
  • llms.txt
  • Newsletter

© 2026 The Open Weights. An independent publication.

PrivacyTermsSMSAggregated by Claude · curated by humans.
LatestNVIDIA2.0
NVIDIAVision-Language

NVIDIA's Nemotron-Parse 2.0 targets document OCR

A compact vision-language model built to turn scanned pages and complex layouts into structured, machine-readable text.

Jun 30, 2026
UpdateOther
Nemotron-Parse-2.0

NVIDIA has published Nemotron-Parse 2.0, a vision-language model aimed squarely at optical character recognition and document parsing. Rather than serving as a general chat assistant, the model is purpose-built to read images of documents and extract their text and structure.

Document parsing sits at the messy intersection of vision and language: forms, invoices, receipts, and scientific papers mix dense text with tables, columns, and figures that trip up naive OCR pipelines. A model tuned for this task is meant to preserve layout and reading order, not just dump characters.

Why it matters

Extracting clean, structured data from documents is one of the most common enterprise uses of AI, and it increasingly feeds retrieval and agent workflows that depend on accurate source text. Specialized parsers like this are attractive because they can be smaller and cheaper to run than routing every page through a large general-purpose model.

A few things to note from the release:

  • It is distributed as a vision-language model on Hugging Face under a custom ("other") license, so teams should review the terms before production use.
  • NVIDIA has not published parameter counts or context details in the record, so evaluation on your own documents remains the best guide.

As the first entry in the Nemotron-Parse line, version 2.0 signals NVIDIA's continued push to broaden its open-weight Nemotron family beyond text generation into practical document intelligence.

Sources

  • nvidia/NVIDIA-Nemotron-Parse-2.0

    Hugging Face

    Visit
OlderMicrosoft previews GELab-Zero-4B, a compact GUI agentMicrosoft · Vision-Language · 5 months agoNewerGigaAI Releases Giga-World-1 Under Apache 2.0Open Gigaai · Image → Video · 5 months ago

Get the model

Hugging Face

Specs

Context window9K tokens
Size3.6 GB
PrecisionFP32
ArchitectureNemotronParseForConditionalGeneration
LicenseOTHER
Downloads589.3K
Likes136

Can you run it?

Runs on just about anything.

  • FP32 (as published)

    8 GB GPU (RTX 4060) · 16 GB laptop or Mac

    5 GB
  • 16-bit

    Any modern laptop · 8 GB RAM

    3.2 GB
  • 8-bit

    Any modern laptop · 8 GB RAM

    2.3 GB
  • 4-bit

    Any modern laptop · 8 GB RAM

    1.9 GB
Your machine
  • KV cache approximated from model size (no config metadata) at 8,192 tokens.

Based on Hugging Face weights metadata. Assumes an 8K context and ~1 GB runtime overhead; actual needs vary. How we estimate


Modalities

Vision-Language

The Weekly Weights

Every open release that mattered, one email a week.

0 comments

No comments yet. Be the first to weigh in.

More from NVIDIA

All NVIDIA releases →
NVIDIA/Text → Video

NVIDIA's Cosmos 3 Brings Open Interactive World Models

The new open release pushes beyond passive video generation toward simulated worlds that respond to input.

Nov 5, 2026
NVIDIA/Image → Video

NVIDIA Releases Cosmos 3 for World Simulation

The new open models push NVIDIA's world-foundation effort toward interactive, real-time video generation.

Nov 5, 2026
Nemotron Labs Dory 1.0 4B
NVIDIA/Text / LLMRuns on a laptop

NVIDIA releases compact Dory 1.0 4B text model

The new Nemotron Labs model packs a 4-billion-parameter language model trained on NVIDIA's Nemotron datasets.

Oct 16, 2026

More in Vision-Language

All Vision-Language →
SenseNova-Loop-9B
SenseTime/Vision-LanguageRuns on a laptop

SenseTime Releases Looped 9B Multimodal Model

The company's new SenseNova-Loop-9B pairs vision-language capabilities with a recurrent looped-transformer design.

Nov 17, 2026
Hunyuan Hy4-flash
Tencent/Vision-LanguageMulti-GPU server

Tencent releases Hunyuan Hy4-flash multimodal MoE

The lightweight, mixture-of-experts model adds vision and text instruction-following to Tencent's open Hunyuan lineup.

Nov 17, 2026
Mirage
Cactus Compute/Vision-Language

Cactus Compute's Mirage brings vision tasks on-device

The new open vision-language model targets detection, segmentation, and visual Q&A without relying on the cloud.

Nov 5, 2026