What are the hardware requirements for Macaw?

Macaw requires approximately 5.4 GB of VRAM and supports a context window of 125K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Macaw and how large is it?

Macaw is developed by badtheorylabs, a model with 2.6b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Macaw?

Macaw is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Macaw by badtheorylabs — VRAM 5.4GB, 125K context

Name: Macaw
Author: badtheorylabs

Macaw is an open-source language model by badtheorylabs. Features: 2.6b LLM, VRAM: 5.4GB, Context: 125K, License: other.

Agent Base model:finetune:liquidai/l... Base model:liquidai/lfm2.5-2.6... Conversational En Endpoints compatible Lfm2 Macos On-device Region:us Safetensors Tool-calling

Model Card on HF 🤗: https://huggingface.co/badtheorylabs/Macaw

Macaw Parameters and Internals

LLM Name	Macaw
Repository 🤗	https://huggingface.co/badtheorylabs/Macaw
Base Model(s)	LiquidAI/LFM2.5-2.6B LiquidAI/LFM2.5-2.6B
Model Size	2.6b
Required VRAM	5.4 GB
Updated	2026-08-09
Maintainer	badtheorylabs
Model Type	lfm2
Model Files	5.4 GB
Supported Languages	en
Model Architecture	Lfm2ForCausalLM
License	other
Context Length	128000
Model Max Length	128000
Transformers Version	5.13.1
Tokenizer Class	TokenizersBackend
Padding Token	<\|pad\|>
Vocabulary Size	128000

Rank the Macaw capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the Macaw

Model	Likes	Downloads	VRAM
Macaw 4bit MLX	14	148	1 GB

Best Alternatives to Macaw

Best Alternatives	Context / RAM	Downloads	Likes
LFM2.5 2.6B	128K / 5.4 GB	81522	398
LFM2.5 2.6B MLX Bf16	128K / 5.3 GB	108	2
LFM2.5 2.6B MLX Mxfp4	128K / 1.6 GB	57	2
LFM2.5 2.6B MLX Nvfp4	128K / 1.6 GB	64	1
LFM2.5 2.6B MLX Mxfp8	128K / 2.8 GB	23	1
LFM2.5 2.6B Base	125K / 5.4 GB	809	37
LFM2 2.6B Exp	125K / 5.1 GB	3203	339
LFM2 2.6B	125K / 5.1 GB	6778	188
LFM2 2.6B Transcript	125K / 5.1 GB	459	171
LFM2 2.6B Transcript ONNX	125K / GB	71	13

Note: green Score (e.g. "73.2") means that the model is better than badtheorylabs/Macaw.