414 posts on this topic
Chinese AI company DeepSeek has released V4.1-Flash-1017, an MIT-licensed open-weight model. Offering a 1 million token context window and a maximum output of 384K tokens, the model raises the bar in coding and agent benchmarks.
With the V4.1-Flash-1017 update to its 552-billion-parameter MoE architecture, DeepSeek lifts its Terminal-Bench 4.0 score to 42.3 and its Humanity's Last Exam result to 42.6, while offering full out-of-the-box optimization for agent frameworks such as Claude Code, DeepSeek Harness and Tencent CodeBuddy.
To simplify the interface of its AI assistant Gemini and put autonomous agent capabilities at the center, Google turned Gems into Skills, Spark into Task mode and Extensions into Apps.
We build, step by step, a type-safe AI Router architecture that classifies incoming user requests with the Jev decision model before sending them to large language models in modern Next.js and TypeScript web applications.
We compare TypeSafe AI Jev and AWS Strands Decider 2B, two key representatives of the decision models becoming the new standard in AI agents, on intelligence score, local execution, calibration and cost.
We take a detailed look at the millisecond latency advantages and zero output token cost of using TypeSafe AI's Jev decision model instead of traditional large language models in production AI workflows.
NVIDIA has made the entire Run:ai platform, which optimizes GPU sharing and AI workload scheduling in enterprise Kubernetes clusters, available as open source on GitHub under the Apache 2.0 license.
We look at how you can solve wrong function calls and broken schemas in autonomous AI agents running in production with the TypeSafe AI Jev model's zero-type-error, calibrated confidence score architecture.
Town, founded by former Plaid CTO Jean-Denis Greze and former Google executive Tony Vincent, reached unicorn status with a $108 million round led by Forerunner Ventures and Menlo Ventures.
We examine the technical details of the TypeSafe AI System One architecture inspired by Daniel Kahneman's cognitive theory, the RLCD training methodology and the Jev model's record 1.6% calibration accuracy.
The US government has issued a sweeping security directive banning products of China-based AI startup DeepSeek, its open-weight models and third-party services built on those models from all civilian federal agencies.
The open-source ArtCraft ecosystem, developed by software engineer Brandon Thomas using Claude AI models and the Rust language, brings subscription-free desktop rivals to Adobe Creative Cloud and Microsoft Office apps.
With the new Chrome extension update released by OpenAI, ChatGPT can analyze open browser tabs to compare products, fill in forms with ChatGPT Work and complete multi-stage workflows autonomously.
Developed by Meta, Muse Spark Open has 816 billion total and 42 billion active parameters and launches with a 1 million token context window, advanced coding skills and local execution on two M5 Ultra Mac Studios with 1 TB of unified memory.
xAI announced Grok Voice 3.0, which offers direct speech-to-speech communication. With low latency and MCP integration, the model plays an active role in Tesla and Starlink operations.
Filing for bankruptcy just five years after its founding, Character.AI has struck a deal with Disney for a $20 million asset sale.
Google's new model Gemini 4 Argon is now available, with benchmark results that beat GPT-6 Astra and Claude Opus 5.5 and a price one-fifth of its rivals'.
Released as open source, WorldPlay2 uses the 14-billion-parameter Wan2.2 architecture and special memory tokens to turn a single image into a consistent virtual world you can explore in real time.
New ChatGPT-based startups are getting ready to shake the business model of giants such as DoorDash, Uber Eats, Yemeksepeti, GetirYemek and Trendyol Go to its core by letting users order directly from restaurants without paying high commissions to intermediary platforms.
Microsoft announced Hybrid Intelligence, an architecture that takes AI in the Windows 11 ecosystem beyond the cloud and brings it together with local hardware. With Copilot's file management, new Search capabilities, the Microsoft Execution Containers security model and RTX Spark powered AI PCs, a new era is beginning for personal computers.
By bringing in former Microsoft Copilot CTO Pedram Rezaei, Amazon strengthened its AWS Agentic AI team and shifted up a gear in the AI agent race.
Google announced Nano Banana 2.1, a model based on Gemini 3.6 Flash. With costs cut in half, improved text rendering and 4K support, it sets a new standard in AI image generation.
Mistral AI crowned Europe's sovereign AI push with the 1-trillion-parameter Mistral Large 4. With an MoE architecture of 49 billion active parameters, a 1M context window and a plan to release open weights, a new era begins in the open-source world.
To meet European Union transparency standards, OpenAI is rolling out textGrain, a system that adds invisible watermarks to ChatGPT and Codex text.
Aleph Alpha announced Kolibri-1, an open-weight model under the Apache 2.0 license that uses only 3.46 billion active parameters per token and has a 1 million token context window.
Ready-made codes to paste into Gemini for a biometric ID photo, a LinkedIn portrait and a simple identity photo. Just replace the square brackets with your own description.
Announced by Tavus, Griffin is a groundbreaking Human Interaction Model (HIM) that understands gestures and facial expressions in real-time video calls and gives human-like instant reactions.
Realistic photo prompts to paste into ChatGPT. In each code, subject, lighting, lens and framing stand apart. That way the result feels like a shot someone took, not a poster.
Ready-made codes for generating portraits, products and outdoor scenes with Nano Banana. In each example, material, lighting and framing are written separately. That way the result doesn't end up looking like a flat filter.
Amazon Web Services (AWS) released Strands Decider 2B, an open-source decision model that minimizes token consumption and latency in AI agents by choosing actions directly instead of generating text.
414 articles about AI, often together with LLM, Technology, OpenAI. The newest one is "Anthropic Unveils Claude Fable 5.2 and Mythos 5.2: Frontier AI for Coding, Science, and Cybersecurity".