Subquadratic tests bolster claims for faster LLM architecture

Miami startup Subquadratic says its SubQ model uses sparse attention to cut computation while preserving strong coding and long-context performance. Independent tests from Appen support parts of the claim, though access remains limited.

Subquadratic has released more evidence for SubQ, a new LLM that it says avoids a long-running bottleneck in transformer-based systems. The Miami startup claims SubQ is faster, cheaper and less energy-intensive than existing models, while processing up to 12 times as much text at once and matching top systems from Google DeepMind, OpenAI and Anthropic on key tasks such as coding.

The company’s approach replaces dense attention, which compares every token with every other token, with sparse attention that selects only some relationships to compute. Subquadratic says its method chooses those relationships dynamically for each input, rather than relying on fixed patterns. Appen, a third-party evaluation firm, found that SubQ was 56 times faster than models using FlashAttention in a speed test and scored 89.7% on LiveCodeBench.

Cost and broader usefulness remain harder to verify because SubQ is not widely available. Dangel said it costs $2600 to run Anthropic’s LLM Opus 4.6 through RULER 128, while SubQ cost eight dollars. Appen also reported that SubQ scored 98% on a needle-in-a-haystack retrieval test with context windows six million and 12 million tokens long.

Skepticism persists because benchmarks do not capture real-world performance, few users have tested the model, and Subquadratic reused weights from a version of the open-source Qwen model rather than training SubQ from scratch. The company says more than 500 enterprise customers are among those seeking early access.

68

Impact Score

Intel jumps as Trump says Apple will partner on U.S. chips

Intel surged after Trump linked Apple to a U.S. chip design and production partnership with the company, potentially strengthening Intel Foundry’s customer story. The rally added to a sharp run, while competitive and valuation concerns remain.

CMA sets new rules for Google search and data access

Google will face new UK requirements covering search ranking transparency and user data transfers. The measures are aimed at giving businesses clearer routes to challenge ranking concerns and more certainty around data-based services.

Trump says Apple will make chips with Intel in the US

Trump publicly claimed Apple agreed to work with Intel on US chip design and manufacturing, but neither company has confirmed the details. The move would align Apple’s search for backup manufacturing with Washington’s push to promote Intel.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.