Prompt Injection

Prompt Injection

Home
Social Media
Imprint
Archive
About
The Hidden Reasoning Hack: Turning Standard Models into Thinking Machines
How we accidentally discovered that some standard LLMs can think like reasoning models - with nothing but a prompt
 • PromptInjection
AI News Roundup: December 13 – December 24, 2025
The most important news and trends
 • PromptInjection
What does AI think about humans? You might be shocked.
A report from a machine’s perspective
 • PromptInjection
StatelessChatUI – One HTML file for direct LLM API access
No installation, no server, no dependencies
 • PromptInjection
When a Local LLM with Q2 Outperforms Cloud Q8: The Parameters Nobody Talks About
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
 • PromptInjection
The 100-Trillion-Token X-Ray: What OpenRouter Reveals About Real AI Usage
Why the loudest debates about AI miss what's actually happening in production
 • PromptInjection
AI News Roundup: December 01 – December 12, 2025
The most important news and trends
 • PromptInjection
The Quantization Cliff: When Does Compression Break Code Intelligence?
Testing GLM 4.5 Air from 4-bit to 1-bit: A Tetris coding experiment reveals where model compression crosses from viable to broken
 • PromptInjection
Llama 4 Scout: Total Disaster – or Misunderstood Workhorse?
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
 • PromptInjection
Can Parameters Compensate for Aggressive Quantization?
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment. A Real-World Test.
 • PromptInjection
AI News Roundup: October 04 – November 30, 2025
The most important news and trends
 • PromptInjection

November 2025

© 2026 Prompt Injection · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture