Posts in AI （719）

* Image is for illustration purposes.

OpenAI o3 Retires from ChatGPT as the 90-Day Notice Period Ends, with No API Changes

NVIDIA Groq 3 LPX Enters Full Production, Hitting 3,400 Tokens per Second at 100K Context

Intel Details Crescent Island Inference GPU: 480GB of Memory in a 350W Air-Cooled Card

Thomson Reuters Unveils Its Own LLM Thomson: 40 Million USD Spent on Legal Specialization, Small Version Released on Hugging Face

* Image is for illustration purposes.

IBM and the USTA Bring New AI Features to the US Open, Scoring Serves 50 Times a Second

* Image is for illustration purposes.

Google Antigravity Now Bundled Into Gemini Enterprise, No Extra License Required

* Image is for illustration purposes.

Antigravity 2.0 Adds Remote Control: Drive Your Machine's AI Agents from Any Browser

OpenAI Publishes First Jalapeño Results: 1.5x to 1.9x Better Performance per Watt

* Image is for illustration purposes.

Adobe Firefly Makes AI Music, Speech and Sound Effects Generally Available, Cleared for Commercial Use

OpenAI Disrupts a Russian Influence Operation Built Around a Fake Think Tank

* Image is for illustration purposes.

Hugging Face Reportedly Exploring a Sale at More Than 13 Billion USD, Roughly Triple Its 2023 Valuation

* Image is for illustration purposes.

Sakana AI to Apply AI to Japan's Ministry of Defense Intelligence Analysis Work

* Image is for illustration purposes.

Liquid AI Releases DSpark Draft Models, Speeding Up LFM2.5 Inference by Up to 3.2x

* Image is for illustration purposes.

Snowflake CoCo Automations Enters Public Preview: Unattended AI Agent Runs Every Hour

NVIDIA to Raise AI Server Prices by More Than 15% as Memory Costs Bite

* Image is for illustration purposes.

Alibaba Launches Wan3.0: 30-Second AI Video Generation That Also Takes Documents

* Image is for illustration purposes.

FreeToken, a MoE-Focused Local Inference Engine, Hits 39 tok/s for a 35B Model on an 8GB GPU

* Image is for illustration purposes.

MCP Publishes a New Roadmap: Long-Running Tasks and Agent Identity Take Top Priority

NVIDIA NVLink Fusion Brings Custom XPUs Into AI Factories, Cutting Latency to One Third

* Image is for illustration purposes.

OpenAI Pauses Its Largest Frontier RL Run While Safeguards Catch Up

* Image is for illustration purposes.

Replit Launches Free Mode, Powered by GPT-5.6 Luna, With No Credit Drain on Everyday Tasks

* Image is for illustration purposes.

Hubble.md Is a Notepad App That Shares the Same Folder With Your AI Agent, Saving Plain Markdown

* Image is for illustration purposes.

Japan's LLM-jp-4 33B Released: A 33.2 Billion Parameter Dense Model Under Apache 2.0

* Image is for illustration purposes.

AWS Makes AgentCore Payments Generally Available, Letting AI Agents Pay for APIs Themselves