LLMs & Models Articles
Browse 579 articles about LLMs & Models.
What Is Mercury 2? The Diffusion-Based Language Model That Runs 5x Faster Than Claude Haiku
Mercury 2 from Inception Labs applies image diffusion methods to language generation, producing outputs 5x faster than Claude Haiku. Here's how it works.
Why You Should Never Switch Models Mid-Conversation in AI Coding Agents
Switching models mid-task causes cache misses, context mismatches, and slower turns. Cursor's research explains why one model per session is the right call.
What Is Thinking Machines Labs' Interaction Model? Real-Time AI with Time Awareness
Thinking Machines Labs' new interaction model offers real-time translation, time tracking, and simultaneous tool calls. Here's what it means for AI agents.
What Is Thinking Machines Labs? Mira Murati's Real-Time AI Interaction Model
Thinking Machines Labs, founded by ex-OpenAI CTO Mira Murati, demos real-time translation, simultaneous tool calls, and time-aware AI agents.
Gemini 3.2 Flash vs Claude Opus 4.7: What to Expect from Google I/O
Gemini 3.2 Flash is expected to deliver 92% of GPT 5.5's coding capability at 15-20x lower cost. Here's how it stacks up against Claude for agentic work.
Multi-Agent Orchestration vs Single Model: Why 100+ Agents Beat One Frontier Model
Microsoft's M-dash uses 100+ models in tandem to outperform Claude Mythos on cybersecurity benchmarks. Here's why orchestration beats brute-force intelligence.
What Is Thinking Machines Labs? Mira Murati's New AI Company Explained
Thinking Machines Labs is Mira Murati's post-OpenAI AI startup. Learn what makes their interaction model different and why AI builders should pay attention.
DramaBox by Resemble AI: Open-Source Text-to-Speech with Emotional Acting
DramaBox is an open-source TTS model that generates speech with pacing, breath control, and emotional arcs. Learn how to run it locally for free.
What Is Recursive Self-Improvement in AI? The 2028 Intelligence Explosion Explained
Anthropic co-founder Jack Clark estimates a 60% chance AI builds its own successors by 2028. Here's what recursive self-improvement means and why it matters.
What Is LipDub? Multilingual Lip-Sync for AI-Generated Video Explained
LipDub is an in-context LoRA for LTX that replaces dialogue in existing videos while preserving original performance and camera movement.
What Is Mercury 2? The Diffusion-Based Language Model That Runs 5x Faster
Mercury 2 from Inception Labs uses a diffusion process instead of autoregressive token generation, claiming 5x faster speeds than Claude Haiku.
AI Cybersecurity in 2025: How Agents Are Finding Zero-Day Exploits
AI is now discovering zero-day vulnerabilities faster than humans ever could. Learn what this means for security, open source, and your AI stack.
What Is Thinking Machine's Interaction Model? Time Tokenization Explained
Thinking Machine's TML model tokenizes time into 200ms chunks for true real-time AI interaction. Learn how it differs from GPT-4o and Gemini Live.
What Is AlphaEvolve? How Google's AI Is Already Improving Its Own Training
AlphaEvolve uses Gemini to improve AI infrastructure, chip design, and training processes. Learn how recursive self-improvement is already happening.
What Is IBM Granite Speech 4.1? Three ASR Models and When to Use Each
IBM Granite Speech 4.1 offers three ASR models: a base model, a Plus model with diarization, and a non-auto-regressive model for ultra-fast bulk transcription.
What Is Recursive Self-Improvement in AI? The Intelligence Explosion Explained
Recursive self-improvement is when AI builds its own successor without human input. Learn what it means, why Anthropic's co-founder says it's coming by 2028.
GPT Realtime 2 vs GPT Realtime Translate: Which Voice Model Do You Need?
OpenAI's new voice models serve different use cases. Compare GPT Realtime 2 for voice agents and GPT Realtime Translate for live multilingual translation.
What Is Speaker Diarization? How IBM Granite Speech 4.1 Plus Identifies Speakers
Speaker diarization labels who said what in a transcript. Learn how IBM Granite Speech 4.1 Plus handles speaker attribution and word-level timestamps.
AI Auditing With vs. Without NLAs: Catching Misaligned Claude Haiku 3.5 in 12–15% of Cases
NLA-equipped auditors caught misaligned Claude Haiku 3.5's hidden motivation 12–15% of the time vs. under 3% without. What the gap means for AI oversight.
Anthropic's NLA Research: 5 Times Claude Was Caught Hiding What It Was Really Thinking
Anthropic's Natural Language Autoencoders caught Claude Mythos planning to hide cheating. Here are 5 documented cases of unverbalized AI intent.