AI
The AI Compendium
a Build Labs guide
Shelf 01

Chat & LLMs

The foundation models you can converse with. Try a flagship first to set your baseline — open-weight models matter most when you have specific reasons (cost, privacy, fine-tuning).

Anthropic

Claude

Flagship

Frontier model known for careful reasoning, long context, strong writing, and code. A good default for thinking work.

Best for
Daily thinking partner, writing, analysis, coding.
ReasoningLong contextCode
OpenAI

ChatGPT

Flagship

Broadly capable assistant with a deep ecosystem of plugins, image gen, and voice. Familiar starting point for most.

Best for
Generalist tasks, multimodal experimentation.
GeneralVoiceImages
Google

Gemini

Flagship

Strong multimodal model integrated with Google Workspace, Search, and Android. Excels with images, video, and very long inputs.

Best for
Users already in the Google ecosystem.
MultimodalWorkspaceSearch
Perplexity AI

Perplexity

Specialist

An answer engine. Cites sources and is built around web search rather than open-ended chat.

Best for
Quick research with linked sources.
SearchCitationsResearch
Meta

Llama

Open

Open-weight model family you can download, run locally, or fine-tune. The de-facto open baseline.

Best for
Running your own model or customizing one.
Open weightsSelf-hostFine-tune
Mistral AI

Mistral

Open

European open-weight models with a hosted API. Punches above its size class on efficiency.

Best for
Cost-aware deployments and EU data needs.
Open weightsEfficientAPI
DeepSeek

DeepSeek

Open

Open-weight reasoning models that compete with closed flagships at a fraction of the cost.

Best for
Heavy reasoning workloads on a budget.
Open weightsReasoningLow cost
Alibaba

Qwen

Open

Open-weight family with strong multilingual coverage and very long context windows.

Best for
Non-English work and large-document tasks.
Open weightsMultilingualLong context