ENPT

PhysicalEye-Decide-35B

Last updated:

Beta51–68ms$0/M input

Quadrant scores

See the full quadrant

Scored with the public quadrant rubric (maturity × capability; bubble = adoption). Revise as evidence lands.

  • Maturity4.8/10

    Apache-2.0 weights (~69 GB bf16, needs an 80 GB GPU) on Darwin-35B-A3B-Opus, a decide.py script, GitHub repo and a live demo Space. No server API or integrations. Small team, no SLA.

  • Capability6.2/10

    Image decisions (up to 4 images) with choice, yes/no and score. Author numbers: egg candling 98% on 100 held-out photos, AI2D 87% on 200 questions, 51 ms median batched on a B200. No ECE, no board result.

  • Adoption1/100

    About 44 downloads and 1 like. No users found.

Vendor claims

Beta: tiny evaluations. The headline numbers rest on 100 egg photos and 200 AI2D questions, run by the author (UnverifiedClaim). No board covers image decisions yet.

Score working (11 Oct 2026)

  • Maturity = 7 × 0.30 + 6 × 0.25 + 3 × 0.25 + 2 × 0.20 = 2.10 + 1.50 + 0.75 + 0.40 = 4.75 → 4.8
  • Capability = 9 × 0.30 + 4 × 0.25 + 6 × 0.25 + 5 × 0.20 = 2.70 + 1.00 + 1.50 + 1.00 = 6.20 → 6.2
  • Adoption = 1 × 0.35 + 1 × 0.35 + 0 × 0.30 = 0.35 + 0.35 + 0 = 0.70 → 1