SubQ – a major breakthrough in LLM intelligence
vanni
20 points
2 comments
May 05, 2026
Related Discussions
Found 5 related stories in 99.4ms across 8,303 title embeddings via pgvector HNSW
- SubQ: a sub-quadratic LLM with 12M-token context mitchwainer · 46 pts · May 05, 2026 · 67% similar
- SubQ: Sub-quadratic LLM built for 12M-token context gagan2020 · 17 pts · May 05, 2026 · 64% similar
- LLMs can unmask pseudonymous users at scale with surprising accuracy Gagarin1917 · 42 pts · March 04, 2026 · 53% similar
- Reliable Software in the LLM Era mempirate · 102 pts · March 12, 2026 · 53% similar
- UK sovereign LLM inference benjamintnorris · 104 pts · May 15, 2026 · 51% similar
Discussion Highlights (1 comments)
Alifatisk
> 12 million token context window > 52x faster than FlashAttention at 1MM tokens > Less than 5% the cost of Opus This sounds too good to be true? Or am I missing something?