What makes a model cheap enough to actually run, including on hardware you own. Quantisation, caching, speculative decoding, and what any of it costs in quality.
FundamentalsDeep dive5 min
Classic LLM pruning commits to one compressed structure before it knows what question a user will ask. TASP calibrates five task-specific masks instead, and a cheap router picks one per turn.
ResearchNews5 min
Explore how the newest generation of reasoning models can now run efficiently at the edge using NVIDIA Jetson, transforming AI capabilities without relying on data centers.
ResearchNews5 min
Understanding NVIDIA's open-source PAIR and its potential to transform local AI processing.
NVIDIA has launched Metatron v1, a model for interactive humanoid control.
MistralAI unveils Timestral-209M, a new model for zero-shot time series forecasting.
NVIDIA has launched the GLM-5.3-Flash-FP4-Local model, a quantized version of ZAI's GLM-5.3-Flash model, optimized for local AI hardware.
Mistral introduces Timestral-123M, a powerful model for zero-shot time series forecasting.
MistralAI has released the Timestral-28M model, a new addition to their time series foundation models.
MistralAI introduces Timestral-5M, a time series foundation model designed for zero-shot forecasting.
Ai2 releases HiRO-ACEv2, a diffusion model for high-resolution climate simulation.
Liquid AI launches LFM2.5-24B-A2B, a 24B-parameter reasoning model designed for fast, private on-device agents.
Liquid AI introduces the LFM2.5-24B-A2B-Base, a 24B-parameter hybrid model optimized for on-device deployment.
NVIDIA releases the GR00T-N1.7-AppleToPlate model with 3 billion parameters.
AWS introduces Claude Haiku 5.5, enhancing speed and efficiency for subagents and high-volume tasks.
Z.ai's GLM 5.3 is now generally available on Amazon Bedrock, offering enhanced capabilities for agentic coding and software engineering.
Dots, GPT-6.1 Sol, Ultrafast, the Agents API and plugins — the announcements that matter from DevDay 2026, and what they mean if you administer IT.
An exploration of OpenAI's new GPT-6 Astra model, its potential costs, performance implications, and expected applications in enterprise environments.
What the new release changes for everyday agent workloads.