Nvidia's Nemotron-Cascade 2 wins math and coding gold medals with 3B active parameters — and its post-training recipe is now open-source

Nvidia's Nemotron-Cascade 2 is a 30B MoE model that activates only 3B parameters at inference time, yet achieved gold ...

Startup Gimlet Labs is solving the AI inference bottleneck in a surprisingly elegant way

Gimlet Labs just raised an $80 million Series A for tech that lets AI run across NVIDIA, AMD, Intel, ARM, Cerebras and ...
Frontiers

Advances in Predictive Coding: Evidence, Mechanisms and Challenges

The study of predictive processing has become a cornerstone in perception science, aiming to explain how the brain anticipates and interprets sensory ...

Mistral's Small 4 consolidates reasoning, vision and coding into one model — at a fraction of the inference cost

Mistral's Small 4 combines reasoning, multimodal analysis and agentic coding in a single open-source model with configurable ...

Figma's Next Phase: AI Monetization and Platform-Wide Expansion

FIG builds a browser-based, AI-powered design hub, expands its suite and AI credits model, while enterprise adoption and ...
The Next Platform

Taalas Etches AI Models Onto Transistors To Rocket Boost Inference

Adding big blocks of SRAM to collections of AI tensor engines, or better still, a waferscale collection of such engines, turbocharges AI inference, as has been shown time and again by AI upstarts ...
TechCrunch

Co-founders behind Reface and Prisma join hands to improve on-device model inference with Mirai

Much of the conversation around AI today is focused on building cloud capacity and massive data centers to run models. Companies like Apple and Qualcomm are in the early stages of making on-device AI ...
winbuzzer.com

AWS Launches SageMaker Inference for Custom Nova Models

Since the Amazon Nova Premier launch and the subsequent Nova customization rollout at AWS NY Summit 2025 brought Nova customization to SageMaker, enterprise customers had been left to bridge one gap ...
Network World

Nvidia claims 10x cost savings with open-source inference models

Nvidia has released analysis showing a 4X to 10X reduction in cost per token for AI inferencing by switching to open source models. The cost discounts required combining Blackwell hardware with two ...
HotHardware

NVIDIA Shows Blackwell Slashing AI Inference Costs By 10X With Open Models

Do you sell AI services? Then NVIDIA wants you to buy Blackwell hardware and host those services yourself, even if you already have perfectly functional Hopper machines. According to NVIDIA, the ...
Ars Technica

OpenAI sidesteps Nvidia with unusually fast coding model on plate-sized chips

On Thursday, OpenAI released its first production AI model to run on non-Nvidia hardware, deploying the new GPT-5.3-Codex-Spark coding model on chips from Cerebras. The model delivers code at more ...
InfoQ

Google BigQuery Adds SQL-Native Managed Inference for Hugging Face Models

Unlock the full InfoQ experience by logging in! Stay updated with your favorite authors and topics, engage with content, and download exclusive resources. This article introduces practical methods for ...