SubQ Snap launches as first long-context decision model: 384K tokens at $0.05/M
rohanpaul_ai · x · 2026-10-13
Subquadratic launched SubQ Snap, billed as the first long-context decision model for enterprise work: constant compute per token as context grows (100K tokens in 660ms, 384K in 2.5s, near-linear scaling), 384K multimodal context, 61.7 on Decision Index v0.2.1 vs Jev's 57.9, and $0.05 per million input tokens with free output. Early 10,000 signups get 100M free input tokens. A shared demo shows agentic repo-search/test workflows cutting cost 21% and latency 18% vs a plain LLM.
More from Models
- Nature Will launches Natura-1, a foundation model for the physical world trained on 1M hours of egocentric video — testingcatalog · 2026-10-13
- User speculates GPT-6.1 Sol's generous limits come from deliberately throttled speed — AccomplishedSugar490 · 2026-10-13
- ChatGPT may have the smallest context window among major chat models, test finds — JeremyNguyenPhD · 2026-10-13
- Patronus tests show prompt-injection detection and action approval are two different tasks — PatronusProtect · 2026-10-13
- DeepMind open-sources 'cryptids': Turing machines with Lean proofs of famous conjectures — gaganghotra_ · 2026-10-13
- Hybrid fine-tuned BERT + LLM pipeline lost its edge to new frontier models — rajistics · 2026-10-13