Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Base model

A base model is a language model straight out of pretraining, before instruction tuning or human-feedback training. It continues whatever text it is given rather than following instructions or holding a conversation.

Also called: pretrained checkpoint, pretrained model, -Base

By Divya Sharma · Checked Nov 15, 2026 · How we review

Ask a base model "What is the capital of France?" and it might answer, or it might write three more quiz questions, because a list of questions is a likely continuation. It has learned language and a lot of facts during pretraining, but it has not been taught to act as an assistant.

Where you see them

Open model releases often ship two versions: a base checkpoint and an instruct model. Some names say so, such as ERNIE 5.5 Flash Base. Hosted APIs almost always serve only the tuned version.

Why you would use one

Base models are the usual starting point for your own fine-tuning, because they have not picked up another team's style, refusals or chat format. Researchers use them to study what pretraining alone produces. They are also good at pure continuation tasks, such as completing code or generating text in a given style.

You can still get useful work out of a base model by framing the prompt as text it would naturally continue, with a few examples first. The GPT-3 paper did exactly this: it tested a model with no fine-tuning, giving tasks and examples only as text. This is few-shot prompting.

How it differs from a foundation model

A foundation model is any broad model meant to be adapted to many tasks. A base model is a specific checkpoint: the untuned one. An instruct model built on it is still a foundation model, but no longer a base model.

Related terms

Instruct model
An instruct model is a base language model that has been further trained on instructions paired with good answers, and often on human feedback, so it follows requests instead of just continuing text.
Pretraining
Pretraining is the first and largest stage of training a model, where it learns general patterns from a very large unlabelled dataset, usually by predicting the next token, before any task-specific tuning.
Foundation model
A foundation model is a large model trained on broad data at scale that can be adapted to many different downstream tasks, through prompting or fine-tuning, instead of being built for one task.
Fine-tuning
Fine-tuning is further training of a pretrained model on a smaller, task-specific dataset, so the model adopts a particular skill, format, tone or domain without being trained from scratch.
Few-shot prompting
Few-shot prompting is giving a language model a handful of worked examples of a task inside the prompt, so it can copy the pattern for a new input without any retraining.

Sources