AI & Models

Frontier language models, how they are trained, and what they can actually do.

13 stories

AI for Hacking: CrowdStrike Says ARTEX Targeted 200+ Orgs

On Oct 29, 2026, CrowdStrike reported that VAPOR PANDA used ARTEX, an AI agent framework running mainly on DeepSeek, to target 200+ organizations in at least 14 countries over 11…

Oct 30, 2026 · 7 min read

AI & Models

Mistral Large 4: What a 1-Trillion-Parameter Open Model Means

Oct 30, 2026 · 8 min read

AI & Models

What Is Gemini 4 Carbon? Google's Bug-Hunting AI, Explained

Oct 29, 2026 · 7 min read

All AI & Models stories

AI & Models

Gemini 4 Carbon: Why Google's New AI Goes to Defenders First

Google announced Carbon on Oct 28, 2026, a security-focused Gemini 4 model that scores 88.2% on SEC-bench Pro and 0.0% prompt-injection success. It goes to a subset of cyber…

Abdul Rehman · Oct 29, 2026 · 6 min read

AI & Models

Open Weight Models Explained: IBM Says 90% as Good, Cheaper

On IBM's Q3 earnings call, reported 2026-10-22, CEO Arvind Krishna said open-weight AI models now cost 10 to 25 percent of frontier token prices and are "90 percent as effective,"…

Abdul Rehman · Oct 27, 2026 · 8 min read

AI & Models

LLM API Pricing Fell Up to 80%: Why, and How to Pay Less

Between June and September 2026, OpenAI cut its small Luna model by 80% and both OpenAI and Anthropic launched cheaper flagships: GPT-6.1 Sol at $2/$10 and Claude Opus 5.5 at…

Abdul Rehman · Oct 26, 2026 · 7 min read

AI & Models

AI Skin Cancer Detection: Why a 10-Year-Old Model Beat GPT-6

Dermatologist Magnus Lynch tested GPT-6 Astra, Claude Fable 5.1 and the 10-year-old Inception v3 CNN on 200 biopsy-confirmed skin lesions. The CNN got 62% of diagnoses right…

Abdul Rehman · Oct 26, 2026 · 8 min read

AI & Models

Nvidia Nemotron: What the Open Model Did and Did Not Prove

In October 2024 Nvidia released Llama-3.1-Nemotron-70B-Instruct, a Llama 3.1 model retuned with a reward model. Its model card reports 85.0 on Arena Hard, ahead of GPT-4o (79.3)…

Abdul Rehman · Oct 25, 2026 · 6 min read

AI & Models

xAI Colossus: What the Memphis Supercomputer Really Is

xAI built Colossus in Memphis with 100,000 Nvidia Hopper GPUs in about 122 days to train Grok. It has since drawn permit disputes over its gas turbines, and reports say Anthropic…

Abdul Rehman · Oct 25, 2026 · 5 min read

AI & Models

BitNet: How Microsoft's 1-Bit LLM Runs AI on a Laptop CPU

Microsoft's BitNet b1.58 stores every model weight as -1, 0 or +1. Its open 2B4T model uses 0.4 GB of memory, and the bitnet.cpp engine runs it on ordinary CPUs faster and with…

Abdul Rehman · Oct 25, 2026 · 7 min read

AI & Models

Speech to Speech AI: Why ChatGPT and Gemini Now Sound Human

Voice assistants such as ChatGPT voice mode and Gemini Live now use single speech to speech models instead of chaining transcription, a text model and a voice reader. OpenAI says…

Abdul Rehman · Oct 25, 2026 · 7 min read

AI & Models

Mixture of Depths: How Google DeepMind Lets AI Skip Layers

Google DeepMind's 2024 mixture-of-depths paper trained transformers whose routers let most tokens skip some layers. Models up to 3B parameters matched or beat standard ones at…

Abdul Rehman · Oct 25, 2026 · 8 min read

AI & Models

Multi-Head Latent Attention: How DeepSeek Shrinks AI Memory

DeepSeek introduced multi-head latent attention in DeepSeek-V2 and kept it in DeepSeek-V3. It compresses the memory a model keeps about earlier text; DeepSeek's paper reports a…

Abdul Rehman · Oct 24, 2026 · 6 min read

All AI & Models stories

The tech that matters, in your inbox.