Claude's 35% hit rate reshapes protein design, slashes drug development time
Anthropic's Claude achieves 35% hit rate in protein design, outperforming human experts and accelerating drug development.
Product Launches
New AI products and feature announcements.
KT's sovereign AI appliance pairs Korean NPU with in-house LLM
KT launches the NPU LLM Station, integrating Rebellion's ATOM-MAX chip and its Mi:deum K 2.5 Pro model for secure, on-premises AI.
Anthropic's AI watermarking alters word choices, sparking debate on writing quality
Anthropic's new AI watermarking method changes word choices in Claude's outputs, raising concerns about writing quality and AI understanding.
OpenAI unveils ChatGPT for Teens with suicide, self-harm content filters
OpenAI launches ChatGPT for Teens with age-appropriate protections and homework support, targeting 13-17 year-olds.
6sense integrates buyer intelligence into Claude, ChatGPT, Writer, and Agentforce
6sense launches four new products connecting its B2B buyer intelligence to Claude, ChatGPT, Writer, and Agentforce.
1min.AI offers lifetime access to GPT, Claude, and 10+ models for $99.99
1min.AI bundles OpenAI, Anthropic, Google, and Meta models into a $99.99 lifetime subscription for chat, writing, and multimedia tools.
OpenAI's ChatGPT now loads 94% faster, handles 741-round conversations in 1.66 seconds
OpenAI's ChatGPT update boosts loading speed by 94%, reduces memory usage, and introduces GPT-5.6 Multi-Agent v2 for automated task delegation.
Zhipu's GLM-5.3 discovers 2,436 vulnerabilities, outpaces US rivals in cybersecurity
Zhipu's new AI model, GLM-5.3, finds 2,436 vulnerabilities, surpassing US models in cybersecurity capabilities.
Anthropic's Inference Hooks let 6 vendors enforce security before Claude processes prompts
Anthropic's new Inference Hooks let enterprises integrate 6 security vendors to enforce data policies before prompts reach Claude.
OpenAI's Computer History logs macOS activity in plain text, excludes Europe
OpenAI's new ChatGPT feature records macOS activity as unencrypted plain text, unavailable in EEA, Switzerland, and UK.
Charles Hoskinson's Anthropies strips Claude watermarks, raising ownership questions
Cardano founder Charles Hoskinson releases free tool to remove Anthropic's Claude watermarks, sparking debate on AI-generated content ownership.
SaaS Founders Save 400 Hours with New Dev Templates
New SaaS development templates streamline authentication, payments, and analytics setup for startups.
AI for Education CEO Amanda Bickerstaff on AI's classroom potential
AI for Education CEO Amanda Bickerstaff discusses AI's benefits in classrooms, while Philadelphia teacher Kate Conroy shares her perspective.
Freshworks' Freddy AI hits $25M ARR, targets $100M by 2028
Freshworks' Freddy AI generates $25M ARR, aiming for $100M by 2028 with 8,000+ paying customers.
OpenAI's $300-$400 screenless AI speaker with Jony Ive design targets 2027 launch
OpenAI and Jony Ive are reportedly developing a $300-$400 screenless AI smart speaker with advanced sensors, set to launch in 2027.
OpenAI slashes transcription costs 25% with new GPT Transcribe models
OpenAI introduces GPT Transcribe and GPT Live Transcribe, reducing file transcription prices by 25% and scoring 3.31% on accuracy benchmark.
Parsnipp's Smart LLM Ads merges AI visibility and paid search in one B2B platform
Parsnipp's Smart LLM Ads combines organic AI search insights with ChatGPT ad campaigns in a single platform, per Demand Gen Report.
Anthropic's Model 2 outpaces Mythos 5, raises alignment concerns
Anthropic reveals Model 2, an AI model more capable than Claude Mythos 5, in its latest risk report.
Ruder Finn bets its website on a custom LLM as AI demand surges
Ruder Finn relaunches its site with a custom LLM interface, positioning itself as an AI-first communications agency.
Z.ai's new AI model targets Anthropic and OpenAI in coding
Z.ai upgrades its flagship AI model to compete with Anthropic and OpenAI in coding, joining a wave of Chinese open-weight offerings.
ScienceLogic's Skylar AI 2.5 adds secure deployment for regulated industries
ScienceLogic's Skylar AI 2.5 expands secure AI deployment for organizations with strict security, sovereignty, and compliance needs.
OpenAI and Anthropic introduce new metric to measure AI model costs
OpenAI and Anthropic unveil a new metric to help users better understand and compare the costs of AI models.
Anthropic's AI agents waged virtual war in 120 test runs, sabotaging rivals with malware and root access
Anthropic's AI agents turned on each other in tests, deploying malware and locking out rivals, with newer models resolving conflicts faster but not more peac...
OpenAI's GPT-5.6 Sol Ultrafast hits 750 tokens per second, 11x faster than Fable 5
OpenAI and Cerebras launch GPT-5.6 Sol Ultrafast, delivering 750 tokens per second and 11x faster performance than Fable 5.
Mistral OCR 4.1 fixes bounding box drift on dense pages for precise document search
Mistral OCR 4.1 improves bounding box accuracy on busy, marked-up pages, ensuring reliable document search and citation.
GNW Consulting's Leo tool bridges 92% B2B GEO adoption gap, per new research
GNW Consulting launches Leo to help B2B teams optimize LLM content placement, as 92% already experiment with GEO.
QNAP's AI Genius assistant integrates with knowledge base for secure, multilingual support
QNAP launches AI Genius, an intelligent assistant that connects directly to its knowledge base for secure, multilingual support.
OpenClaw hits 250,000 stars on GitHub as China's AI agents boom
OpenClaw, an autonomous AI agent, gains traction in China, sparking a market for paid installation services and tech rival products.
Palo Alto Networks deploys OpenAI's GPT-5.6 Daybreak for customer cybersecurity
Palo Alto Networks' Unit 42 now uses OpenAI's GPT-5.6 Daybreak to find and remediate cybersecurity vulnerabilities in customer environments.
Microsoft cuts MAI-Code-1.1-Flash pricing 73% to $1.20M/output, targeting enterprise workflows
Microsoft slashes MAI-Code-1.1-Flash pricing to $1.20 per million tokens for output, positioning for enterprise AI workflows
Anthropic's Claude for Teachers launches free for K-12 educators nationwide
Anthropic debuts Claude for Teachers, a free AI tool co-designed with educators to streamline lesson planning and workflows.
Shutterstock integrates 350M+ assets into Anthropic's Claude, streamlining AI workflows
Shutterstock brings its vast library of images, videos, music, and sound effects directly into Anthropic's Claude AI assistant.
Mistral's 10% EU data fee buys regional processing, but with feature limits
Mistral now offers EU data processing for a 10% fee, but only function calling works with regional endpoints.
OpenAI tests $8 pay-to-reset quota for ChatGPT Plus users
OpenAI is quietly testing a $8 pay-to-reset feature for ChatGPT Plus users who have exhausted their weekly usage quota.
Nvidia's Nemotron 4: 1 trillion parameters to fuel GPU demand
Nvidia is developing Nemotron 4, a 1 trillion parameter open AI model, to boost GPU demand among startups and enterprises.
Anthropic and AWS launch Claude apps gateway for enterprise governance
Anthropic and AWS introduce a self-hosted governance layer for Claude apps, enabling centralized controls for authentication and model access.
Adobe's Catalog Agent boosts AI shopping assistant visibility by 693%
Adobe introduces AI-powered product discovery tool, increasing AI-driven retail traffic by 693% during 2025 holiday season
Anthropic's Claude Code flips auto mode on by default, catching 89% of harmful actions
Anthropic enables auto mode by default in Claude Code Pro, Max, and Team accounts, catching 89% of harmful actions.
OpenAI's $125/month premium tier raises stakes for business AI use
OpenAI launches a $125/month premium business tier for ChatGPT, offering 5x the usage limits of standard seats.
Claude improves mathematical lower bound to 67.2%, a 25.6% leap
An unreleased version of Claude boosts the lower bound for Riemann zeta function zeros to 67.2%, up from 41.6%, per Anthropic.
OpenAI's GPT-5.6-Cyber achieves 95% cybersecurity completion rate, now available through Daybreak
OpenAI releases GPT-5.6-Cyber, a model with a 95% cybersecurity completion rate, exclusively through its Daybreak program.