Read More: The fight against climate change is entering a challenging phase: it is no longer enough to simply reduce pollution, and scientists are talking about removing up to 9,700 million metric tons of CO2 per year by 2050<\/a><\/h3>\n<\/div>\n<\/div><\/div>\n<\/div>\n\n\n\nThe trouble starts when the text gets very long. Standard attention compares many word pieces with many other word pieces, and the amount of work grows fast. Picture every student in a packed gym having to compare notes with every other student before answering one question.<\/p>\n\n\n\n
What SubQ says it changed<\/h2>\n\n\n\n
Subquadratic says SubQ uses a method called sparse attention<\/a>. In simple terms, the model does not compare every token with every other token. It tries to focus only on the relationships that matter most.<\/p>\n\n\n\nThat sounds obvious, right? But it has been hard to do without making the model worse. Alex Whedon, co-founder and chief technology officer of the startup, has argued that language is too complex for fixed shortcuts, so SubQ chooses important relationships dynamically for each input.<\/p>\n\n\n\n
In practical terms, that could matter for jobs that involve huge files. A lawyer might want to search many contracts at once<\/a>. A developer might want an AI tool to inspect an entire codebase instead of small slices of it.<\/p>\n\n\n\nBenchmarks raised the stakes<\/h2>\n\n\n\n
The outside evaluation reported strong results on long-context retrieval, which means finding a specific fact buried inside a large amount of text. The evaluator said SubQ 1.1 Small Preview returned exact answers every time at one million and two million tokens, and reached 98 percent exact-match accuracy at six million and 12 million tokens.<\/p>\n\n\n\n
On LiveCodeBench, a coding benchmark that collects new programming problems over time to reduce the risk that models have already seen the answers, SubQ reached 89.7 percent pass at four attempts across more than 1,000 problems. That means the model got credit when at least one of four tries solved the task.<\/p>\n\n\n\n
Subquadratic also says SubQ 1.1 Small uses 64.5 times less compute than dense attention at one million tokens and runs 56 times faster than FlashAttention-2<\/a> on a single attention layer. Those are large claims, and they are exactly why the AI community is watching closely.<\/p>\n\n\n\n![\"A](\"https:\/\/www.ecoticias.com\/en\/wp-content\/uploads\/2026\/07\/subquadratic-subq-ai-model-long-context-1.jpg\" "\\"\\"")
Miami startup Subquadratic is drawing significant industry attention with SubQ, an AI model that claims to process massive document stacks without the traditional speed and cost bottlenecks.<\/figcaption><\/figure>\n\n\n\nWhy this is not settled yet<\/h2>\n\n\n\n
Benchmarks are useful, but they are not the same as everyday use. The developers of RULER, a long-context testing suite from Nvidia, note that their benchmark is not comprehensive and should not replace realistic tasks.<\/p>\n\n\n\n
There is another catch. SubQ is still in limited access, so most outside developers cannot test it on messy real-world work. That means questions remain about reliability, edge cases, and how it behaves when documents are incomplete, duplicated, or full of contradictions.<\/p>\n\n\n\n
That sounds obvious, right? But it has been hard to do without making the model worse. Alex Whedon, co-founder and chief technology officer of the startup, has argued that language is too complex for fixed shortcuts, so SubQ chooses important relationships dynamically for each input.<\/p>\n\n\n\n
In practical terms, that could matter for jobs that involve huge files. A lawyer might want to search many contracts at once<\/a>. A developer might want an AI tool to inspect an entire codebase instead of small slices of it.<\/p>\n\n\n\n The outside evaluation reported strong results on long-context retrieval, which means finding a specific fact buried inside a large amount of text. The evaluator said SubQ 1.1 Small Preview returned exact answers every time at one million and two million tokens, and reached 98 percent exact-match accuracy at six million and 12 million tokens.<\/p>\n\n\n\n On LiveCodeBench, a coding benchmark that collects new programming problems over time to reduce the risk that models have already seen the answers, SubQ reached 89.7 percent pass at four attempts across more than 1,000 problems. That means the model got credit when at least one of four tries solved the task.<\/p>\n\n\n\n Subquadratic also says SubQ 1.1 Small uses 64.5 times less compute than dense attention at one million tokens and runs 56 times faster than FlashAttention-2<\/a> on a single attention layer. Those are large claims, and they are exactly why the AI community is watching closely.<\/p>\n\n\n\n Benchmarks are useful, but they are not the same as everyday use. The developers of RULER, a long-context testing suite from Nvidia, note that their benchmark is not comprehensive and should not replace realistic tasks.<\/p>\n\n\n\n There is another catch. SubQ is still in limited access, so most outside developers cannot test it on messy real-world work. That means questions remain about reliability, edge cases, and how it behaves when documents are incomplete, duplicated, or full of contradictions.<\/p>\n\n\n\nBenchmarks raised the stakes<\/h2>\n\n\n\n
Why this is not settled yet<\/h2>\n\n\n\n