DigiNews

Tech Watch by Johan Denoyer

← Back to articles

Introducing SubQ 1.1 Small

Quality: 8/10 Relevance: 9/10

Summary

SubQ 1.1 Small release introduces Subquadratic Sparse Attention (SSA) to remove quadratic context constraints, enabling long-context reasoning up to 12M tokens with large compute savings. The post covers features, benchmarks, efficiency, training, and use cases, with plans for broader model rollout and partner deployments.

🚀 Service construit par Johan Denoyer