ModelLLMsv2.0

Mamba 2

by Carnegie Mellon / Princeton · open-source · Last verified 2026-03-17

Second-generation selective state space model achieving transformer-competitive quality with linear-time sequence processing. Introduces structured state space duality (SSD) for 2-8x faster training throughput compared to the original Mamba architecture.

https://github.com/state-spaces/mamba ↗

C—Below Average

Adoption: DQuality: BFreshness: C+Citations: C+Engagement: F

Specifications

License: Apache 2.0
Pricing: open-source
Capabilities: text-generation, linear-time-inference, efficient-training, long-sequence-processing, state-space-modeling
Integrations: huggingface, transformers
Use Cases: efficient-inference, long-sequence-modeling, research, edge-deployment
API Available: No
Parameters: 2.7B
Context Window: Unlimited (state space)
Modalities: text
Training Cutoff: Early 2024
Tags: llm, open-source, state-space-model, linear-complexity, efficient
Added: 2026-03-17
Completeness: 82%

Index Score

40.15

Adoption

Quality

Freshness

Citations

Engagement

Explore the full AI ecosystem on Agents as a Service