Bright

BRIGHT WIRE / OPEN MODELS

Open models

Open weights, repositories and the systems people build from them.

Topic RSS · Topic JSON Feed · Topic API · Latest-item embed

Published records

Emerging · Bright published 2026-10-04

Your AI conversation feels private. Where it runs matters.

A chatbot can make it easy to think out loud. The harder question is who else can access those thoughts. For people exploring personal questions, working through an unfinished idea or handling confidential material, where an AI runs is a practical privacy choice.

Source published 2026-09-30 · Evidence pack

Emerging · Bright published 2026-09-28

NVIDIA adds an independent safety layer for AI agents.

AI agents can write code, use tools and keep working long after a person steps away. NVIDIA’s September 28 Open Agent Safety Platform launch puts permission controls outside the agent’s workload and offers a separate hardware-backed watchdog. The aim is to give people stronger control over what agents can reach; the launch does not establish that agents are now safe in every setting.

Source published 2026-09-28 · Evidence pack

Experimental · Bright published 2026-09-19

A local visual-inspection workbench

Anomalib packages multiple anomaly-detection methods into a modular library for finding and localizing unusual regions in inspection images, with paths from experiments to optimized edge inference.

Source published 2022-02-16 · Evidence pack

Emerging · Bright published 2026-09-19

Assembling biological model workflows

BioNeMo brings biological foundation models and training components into one framework for protein, molecular, and drug-research experiments.

Source published 2024-11-15 · Evidence pack

Experimental · Bright published 2026-09-19

Simulating a scene before a robot enters it

Cosmos 3 provides downloadable omnimodal world-model variants that can process or generate combinations of text, images, video, audio, and action sequences for physical-AI research.

Source published 2026-05-31 · Evidence pack

Demonstrated · Bright published 2026-09-19

Predicting a protein’s shape from its sequence

ESMFold applies a protein language model to infer three-dimensional structure directly from an amino-acid sequence, with checkpoints and bulk prediction tools available to researchers.

Source published 2023-03-16 · Evidence pack

Emerging · Bright published 2026-09-19

A security model for a SOC's own evidence

Foundation-Sec-8B is a cybersecurity-focused Llama 3.1 derivative that organizations can download and adapt for security operations work involving their own alerts, cases, and threat knowledge.

Source published 2025-04-28 · Evidence pack

Emerging · Bright published 2026-09-19

An open model learns the Moon's surface.

NASA and IBM released a Lunar Foundation Model trained primarily on Lunar Reconnaissance Orbiter data for research tasks such as crater and volcanic-feature mapping and possible polar-ice analysis.

Source published 2026-09-10 · Evidence pack

Demonstrated · Bright published 2026-09-19

Proposing materials for a desired property

MatterGen generates candidate inorganic crystal structures while conditioning on properties a researcher wants, changing the starting point from searching a known catalog to proposing structures for testing.

Source published 2025-01-16 · Evidence pack

Experimental · Bright published 2026-09-19

Sketching music from words and melody

MusicGen generates music from a text description and can use an input melody as an additional condition, providing a downloadable tool for research and creative prototyping.

Source published 2023-06-08 · Evidence pack

Demonstrated · Bright published 2026-09-19

Adapting one robot policy across nine platforms

Octo is a generalist manipulation policy trained on the Open X-Embodiment dataset and evaluated as a reusable starting point for robots with different sensors, action spaces, and physical forms.

Source published 2024-05-20 · Evidence pack

Experimental · Bright published 2026-09-19

Teaching a robot from demonstrations

OpenVLA gives robotics teams a pretrained vision-language-action model plus instructions for fine-tuning it on their own robot demonstrations.

Source published 2024-06-13 · Evidence pack

Demonstrated · Bright published 2026-09-19

Adapting one Earth model to many maps

Prithvi-EO-2.0 is a geospatial foundation model that can be adapted to Earth-observation tasks such as flood mapping, crop classification, and land-use analysis.

Source published 2024-12-03 · Evidence pack

Experimental · Bright published 2026-09-19

An Earth model runs above the Earth.

A compressed NASA–IBM Prithvi geospatial model ran on the Kanyini satellite and the ISS IMAGIN-e payload in flood and cloud detection demonstrations.

Source published 2026-05-07 · Evidence pack

Demonstrated · Bright published 2026-09-19

SAM 1: selecting an object with a point or a box

The original 2023 Segment Anything release, now commonly called SAM 1, turns a click, box, or rough mask into an object segmentation, giving annotators and analysts an editable first pass across varied images.

Source published 2023-04-05 · Evidence pack

Demonstrated · Bright published 2026-09-19

Adapting an image model to a visual practice

Stable Diffusion's downloadable latent-diffusion weights and adaptation code allowed artists and developers to build local image generation, inpainting, fine-tuning, and LoRA workflows.

Source published 2022-08-22 · Evidence pack

Demonstrated · Bright published 2026-09-19

A year of satellite observation, made lighter to carry.

TESSERA provides 10-meter, pixel-wise annual embeddings learned from Copernicus Sentinel-1 and Sentinel-2. ESA says the model first launched in 2025; its repository and a peer-reviewed CVPR 2026 paper now document the work.

Source published 2026-06-04 · Evidence pack

Demonstrated · Bright published 2026-09-19

Transcribing speech on infrastructure you control

Whisper provides downloadable speech-recognition models for transcription, language identification, translation into English, and caption-making without requiring a hosted speech service.

Source published 2022-09-21 · Evidence pack

Deployed · Bright published 2026-09-09

A world-class forecast now fits on one machine.

Weather forecasting has spent fifty years as one of the largest users of supercomputing. AI models trained on the archive now produce comparable global forecasts in minutes, and the weights are published. The measurements they depend on are the part that has not become cheap.

Source published 2025-12-17 · Evidence pack

Deployed · Bright published 2026-09-07

Reasoning weights people can run themselves.

OpenAI released gpt-oss-120b and gpt-oss-20b as Apache-2.0 open-weight reasoning models, with downloadable weights and reference implementations for local deployment.

Source published 2025-08-05 · Evidence pack

Deployed · Bright published 2026-09-07

An open reasoning model with its recipe beside it.

NVIDIA released Nemotron 3 Super, an open 120-billion-parameter mixture-of-experts reasoning model with 12 billion active parameters, plus stated releases of its methodology, data, reinforcement-learning environments, and evaluation recipes.

Source published 2026-03-11 · Evidence pack

Demonstrated · Bright published 2026-09-07

A model whose full recipe is visible.

Ai2 released OLMo 2 32B, a language model accompanied by publicly available data, code, weights, training details, and a reproducible training recipe.

Source published 2025-03-13 · Evidence pack

Deployed · Bright published 2026-09-07

Choosing when a model thinks longer.

The Qwen team released and described Qwen3, a family of dense and mixture-of-experts language models with selectable thinking and non-thinking modes.

Source published 2025-04-29 · Evidence pack