Sarvam 105B, the first competitive Indian open source LLM

LLMs work best when the user defines their acceptance criteria first

A tool that removes censorship from open-weight LLMs

Labor market impacts of AI: A new measure and early evidence

A standard protocol to handle and discard low-effort, AI-Generated pull requests

GPT 5.4 Thinking and Pro

The L in "LLM" Stands for Lying

Something is afoot in the land of Qwen

Qwen3.5 Fine-Tuning Guide – Unsloth Documentation

LLMs can unmask pseudonymous users at scale with surprising accuracy

GPT‑5.3 Instant

Claude is an Electron App because we've lost native

Gemini 3.1 Flash-Lite: Built for intelligence at scale

Claude's Cycles [pdf]

Language Model Contains Personality Subnetworks

Right-sizes LLM models to your system's RAM, CPU, and GPU

When does MCP make sense vs CLI?

Why XML Tags Are So Fundamental to Claude

Microgpt explained interactively

10-202: Introduction to Modern AI (CMU)

Microgpt

Running a One Trillion-Parameter LLM Locally on AMD Ryzen AI Max+ Cluster

Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers

Unsloth Dynamic 2.0 GGUFs

We gave terabytes of CI logs to an LLM

I don't know how you get here from “predict the next word”

Bcachefs creator insists his custom LLM is female and 'fully conscious'

I asked Claude for 37,500 random names, and it can't stop saying Marcus

AIs can't stop recommending nuclear strikes in war game simulations

Mercury 2: The fastest reasoning LLM, powered by diffusion

Making Wolfram Tech Available as a Foundation Tool for LLM Systems

"Car Wash" test with 53 models

AIs can generate near-verbatim copies of novels from training data

Anthropic Education the AI Fluency Index

How Taalas “prints” LLM onto a chip?

Minions – Stripe's Coding Agents Part 2

The path to ubiquitous AI (17k tokens/sec)

Fast KV Compaction via Attention Matching

Consistency diffusion language models: Up to 14x faster, no quality loss

Gemini 3.1 Pro

What years of production-grade concurrency teaches us about building AI agents

Study: Self-generated Agent Skills are useless

Editor's Note: Retraction of article containing fabricated quotations

Two different tricks for fast LLM inference

ai;dr

Improving 15 LLMs at Coding in One Afternoon. Only the Harness Changed

GPT-5 outperforms federal judges in legal reasoning experiment

Mathematicians issue a major challenge to AI—show us your work

GPT-5.3-Codex being routed to GPT-5.2

RLHF from Scratch