ModelLLMsv1.0

MegalodonLM

by Meta FAIR · free · Last verified 2026-03-17

MegalodonLM is a research model from Meta FAIR introducing the Megalodon architecture, a novel efficient sequence modeling approach that achieves Transformer-level quality with linear complexity scaling. It represents a fundamental architectural advance in long-context language modeling without quadratic attention costs.

https://github.com/XuezheMax/megalodon ↗

D—Poor

Adoption: FQuality: B+Freshness: B+Citations: DEngagement: F

Specifications

License: Apache 2.0
Pricing: free
Capabilities: text-generation, long-context, efficient-inference, research
Integrations: Hugging Face
Use Cases: long-sequence-modeling, research, efficient-language-modeling, architecture-exploration
API Available: No
Parameters: 7B
Context Window: 4K
Modalities: text
Training Cutoff: 2023
Tags: meta, research, linear-attention, efficient, ssm
Added: 2026-03-17
Completeness: 95%

Index Score

33.1

Adoption

Quality

Freshness

Citations

Engagement

Explore the full AI ecosystem on Agents as a Service