AIHow AI Works

Tag: LLM

What One AI Query Costs: Tokens, Bytes, Joules, Cents

One question, one model, one machine — costed end to end. Then four dials that move the same answer by more than 10,000x.

Karpathy’s nanochat and LLM Council: Build Your Own AI

nanochat is Andrej Karpathy's ~$100 from-scratch ChatGPT clone, and LLM Council polls several frontier models. What each is and how they work.

How to Build an Agent: Thorsten Ball’s 400-Line Blueprint

Thorsten Ball's 'How to Build an Agent' shows a code-editing AI is just an LLM in a loop with three tools, in under 400 lines of Go.

MiniMax M3 Open-Sourced on Hugging Face

MiniMax has open-sourced its M3 model, a 428B MoE architecture optimized for long context and agentic scenarios, now available on Hugging Face.

LEAP: How LLMs Solved Every Putnam 2025 Problem

Google's LEAP paper details a system for automated theorem proving. It lets LLMs write formal Lean proofs and has solved 12 of 12 Putnam 2025 problems.

DeepSeek-V4: Pro and Flash Models with 1M Context

DeepSeek releases V4-Pro and V4-Flash models featuring 1.6T parameters, open-source weights, and a massive 1 million token context window.

xAI Launches Grok Voice Think Fast 1.0

xAI announces Grok Voice Think Fast 1.0, a new flagship voice model built for complex workflows, real-time reasoning, and zero added latency.

Xiaomi MiMo-V2.5: The Next Generation of Open Agentic Models

Xiaomi announces the MiMo-V2.5 series, featuring flagship agentic performance and native omnimodality to rival frontier models like Claude 4.6 and GPT-5.4.

Kimi K2.6 Release: Open Weights and 12-Hour Long-Horizon Coding

Moonshot AI releases Kimi K2.6, featuring open weights, impressive coding benchmarks, and support for agentic swarms with up to 300 sub-agents.

Google Gemma 4: The Next Frontier of Open Models for AI Agents

Google introduces Gemma 4, a new family of open models optimized for complex reasoning, autonomous agents, and tool use with up to 256K context window.