Company Terminal

PrismML

PrismML is an AI model lab, spun out of Caltech, that builds highly compressed open-weight language models — the Bonsai family — designed to run on edge devices such as phones and laptops rather than requiring cloud servers. Its core technique is storing model weights in 1-bit or ternary form to maximize capability per bit.

Momentum
Not scored
as of 2026-08-03
Funding
—
Founded
—
Stack
Layer 4
Research Tier A
0%
needs citations
Momentum inputsNot scored · need ≥2 signals · as of 2026-08-03

No momentum signals yet for this company. See docs/MOMENTUM.md.

No radar series.

Linked model value

    No headcount (est.) series yet.
    No capital raised (est.) series yet.
    No monthly web attention (est.) series yet.
    No geographic reach (countries, est.) series yet.

    Research completeness

    • Tier A: 0% (0/0)
    • Tier B: 0% · Tier C: 0%
    • Publish gate: Blocked
    Open in Research Desk →

    Citations / claims

    No claims yet. Run seed_provenance_tier_a.sql or add sources in Research Desk.

    What they do

    Who is this for? Teams evaluating AI vendors in this category.

    Why different? PrismML is an AI model lab, spun out of Caltech, that builds highly compressed open-weight language models — the Bonsai family — designed to run on edge devices such as phones and laptops rather than requiring cloud servers. Its core technique is storing model weights in 1-bit or ternary form to maximize capability per bit.

    Models & technologies

    Competitors

    Provenance

    Profile loaded from supabase. Claims and metric observations carry confidence labels; prefer primary URLs in Research Desk over estimate backfills.