Runway releases Praxis‑1 world action model with open weights
Runway has made Praxis‑1, a 27-billion-parameter model designed to predict robot video and actions jointly, generally available with Apache 2.0 open weights.
1 min readRead story
CHOOSE YOUR FOCUS
What do you create?
NEWS FEED
Open weights
TopicOpen weightsClear topic
Sorted by AI Radar publication time · Times shown in Istanbul (UTC+3).Clear filters
NVIDIA releases Alpamayo 2 Edge weights for onboard autonomous driving
The two-billion-parameter Alpamayo 2 Edge model is available with downloadable weights and DRIVE AGX Thor recipes for generating driving rationale and future trajectories within onboard compute and latency limits.
2 min readRead story
Thinking Machines makes Inkling 2 Small available on Tinker
Inkling 2 Small uses a 594B-total, 17B-active MoE architecture with image input and long-horizon agent capabilities; it is available today on Tinker with closed weights.
2 min readRead story
Baidu releases open weights for ERNIE 5.5 Flash Base
ERNIE 5.5 Flash Base has 280 billion total parameters with 14 billion activated per token, and supports a 1-million-token context for fine-tuning and inference.
2 min readRead story
IBM Granite Switch 4.1 combines chat, decisions, safety, and speech on one endpoint
Granite Switch 4.1’s 8B and 30B models run chat, decision-making, safety checks, and English speech recognition on one endpoint with a shared KV cache.
2 min readRead story
Meta releases weights for its 816B-parameter Muse Spark Open model
Muse Spark Open combines 816 billion total and 42 billion active parameters with a one-million-token context window, accepting text, images, audio, and video as input.
2 min readRead story
Reflection releases weights and agent tools for its 501B-parameter Beam model
Beam’s Apache 2.0-licensed BF16, FP8, and NVFP4 weights are now on Hugging Face, connecting the model to coding and long-horizon agent work through Mirror CLI and widely used inference frameworks.
2 min readRead story
Black Forest Labs releases the 5.6-billion-parameter FLUX 3 Image Klein
The open-weight model brings image generation, editing, and combinations of up to 10 references to smaller local hardware.
2 min readRead story
Musubi PolicyLM evaluates content rules with a compact model
The 1.7-billion-parameter open-weight model scores text against supplied policy categories. It is not a chatbot generating free-form explanations.
1 min readRead story
NVIDIA explains Nemotron’s olympiad-level training recipe
Nemotron specializations exceeded gold thresholds in coding and mathematics. NVIDIA explains the models, data and generate-verify-refine workflow.
1 min readRead story
BEYOND THE HEADLINE
What changed?
What can it bring to your work?
Every story includes a concise summary, its creative workflow implications and original sources. Explore official demos and examples alongside the news.
Our editorial approach
