SubQ launches Snap: a 384K-context decision model at $0.05 per million input tokens
ryanallred · x · 2026-10-13
SubQ introduced Snap, billed as the first long-context "decision model" for enterprise work:
- 384K multimodal context, 12x larger than Jev
- Compute per token stays constant as context grows
- Scores 61.7 on Decision Index v0.2.1 vs Jev's 57.9, winning 13 of 16 enterprise benchmarks
- Returns a decision on 100K tokens in under one second
- $0.05 per million input tokens, output free
The pitch: dedicated compute for the routing/filtering/classifying parts of agent workflows that never needed reasoning prices. First 10,000 sign-ups get 100M free input tokens.
More from coding & agent
- Ten AI agents, one email: users mute agents after notification overload — annbordetsky · 2026-10-13
- Blog post shrinks from 158KB HTML to 17KB Markdown for AI agents — colinmcnamara · 2026-10-13
- Amp's diff view now auto-orders changed files by relevance, 10-30x faster — HankYeomans · 2026-10-13
- AI Engineer New York opens with a print daily: 300K-dev telemetry says org, not model — jonmarkgo · 2026-10-13
- One-line decorator turns any combinatorial solver into a differentiable NN layer — PMinervini · 2026-10-13
- Google's enterprise Gemini agent: one memory across Workspace, M365 and Slack, routes work to Claude — BenBajarin · 2026-10-13