About this episode

The episode discusses Subquadratic's innovative Sparse Attention architecture and its implications for long-context performance in AI models.

Subquadratic is beginning to back up its ambitious claims with benchmarks and third-party validation for its SubQ 1.1 Small model, which uses its proprietary Sparse Attention (SSA) architecture to dramatically improve long-context performance. Rather than comparing every token to every other token, SSA selectively processes relationships, enabling near-linear scaling while maintaining high accuracy across context windows of up to 12 million tokens. The company reports near-perfect retrieval performance, competitive coding and reasoning benchmarks, and compute savings of up to 1,000x at maximum context lengths.

Topics covered

• technology
• artificial intelligence
• machine learning
• performance benchmarks
• context processing

Keywords

• Subquadratic
• Sparse Attention
• long-context performance
• AI benchmarks
• compute savings

Mentioned in this episode

Organizations: Subquadratic

Products: SubQ 1.1 Small, Sparse Attention (SSA)

More episodes of The New Stack Podcast

• Meet Brain, the AI that decides when Azure is officially down · July 14, 2026 · 19 min
• “The harness is where the hard work is”: Harness bets on agents that enterprises can trust in production · July 2, 2026 · 20 min
• Public cloud vs. on-prem: Summit on where each workload belongs · June 25, 2026 · 37 min
• Gusto Cofounder: An AI agent that runs payroll, HR, and benefits without waiting to be asked · June 18, 2026 · 29 min

Explore listener stats, chart rankings, contacts and more on the The New Stack Podcast podcast page.