Fable 5.1 lands in Claude Code: longer context, cheaper tokens
· One min read
Anthropic just released Claude Fable 5.1 and promoted it to the default model in Claude Code. Three changes matter if you write code every day.
6 posts tagged with "LLMs"
Large language models: how they work, what they cost, and where they fail.
View All TagsMCP one year in: the three servers we install by default
· One min read
The Model Context Protocol (MCP) has been in production for a year. We now install three servers by default in any serious repo that uses Claude Code.
Tokens, Context Windows and What They Really Cost
· 7 min read
Two facts explain most surprising invoices. A token is roughly ¾ of an English word, so 1,000 tokens is about 750 words. And a conversation has no memory — every turn resends the entire history, so turn twenty costs many times what turn one did.
Why LLMs Hallucinate, and What Actually Reduces It
· 7 min read
The uncomfortable truth first: a language model producing a fabricated citation is doing exactly the same thing as when it produces a correct one. There is no separate "making things up" mode that switches on. Understanding that is the difference between chasing the problem forever and actually reducing it.
What GPU Do You Need to Run an LLM Locally?
· 7 min read
The rule that answers most of the question: a model needs roughly its parameter count multiplied by the bytes per parameter, plus about 20% overhead. A 7-billion-parameter model at 16-bit precision is 7 × 2 = 14 GB of weights, so around 17 GB of VRAM in practice. Quantize it to 4-bit and the same model fits in about 5 GB.
RAG vs Fine-tuning: How to Actually Choose
· 8 min read
The short answer, and it is worth stating before anything else: RAG changes what the model knows, fine-tuning changes how the model behaves. Most teams reach for fine-tuning when they actually have a knowledge problem, spend six weeks on it, and end up with a model that sounds right and gets facts wrong.