SubQ 1M-Preview

Released
2026-05-05
Last refreshed
2026-10-13
Status
Researched 160d ago
ProprietaryCommercial use: restrictedRAGAgentsLong contextJSON / Tool use
Subquadratic releases · 3 in the last 12 months · this family litChangelog →
Specifications
Family
SubQ
Released
2026-05-05
Context
1M
Architecture
Decoder Only
Specialization
general
Openness
Proprietary
License
ProprietaryRestricted
Weights
Not released
Code
Unknown
Training
Pretrained
Created by
Subquadratic

Efficiency is Intelligence

Founded 2026
Website
Pricing

No tracked provider pricing is available yet.

Providers(1)
SubQ API
Links
Website

About

SubQ 1M-Preview is Subquadratic's sub-quadratic sparse-attention LLM: linear O(n) attention versus O(n²) standard transformers, with 1M production context tested to 12M. Subquadratic claims roughly 50× faster/cheaper inference at long context—vendor-claimed, not independently verified. The model posts vendor-claimed SWE-bench Verified 81.8%, RULER 95.0%, and MRCR 65.9% in lab prose; LLMReference has no independent benchmark rows yet. OpenAI-compatible API access is available via SubQ AI.

Providers(1)

Compare API pricing across 1 tracked provider route for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MRoute
SubQ API--
ServerlessPartial

Capabilities

ReasoningJSON / Tool use

Benchmark peer barsfor RAG

No task-mapped benchmark peers are available for this model yet.

Migration checks

No linked migration route is available for this model yet.