Skip to content
VECTOR WIREAI INTELLIGENCE
UTC
Pro Mode — raw scores, one benchmark under the lensthird-party eval harness
The field · 10 measured
24.0964.3

10 models measured, most on the third-party eval harness. Mage-VL-4B tops the board at 64.3.

10 measured·Lifecycle Updated Oct 8, 2026Score basis
#◆Vendor$/M in$/M out
Leader — ignites whiteField — fill cools with scoreBars show 24.09–64.3 · ◆ solid = first-party or better · outlined = arm's-length
01Mage-VL-4BMicrosoft64.3———
02HY-Embodied 0.5 MoT-2BTencent60.5VENDOR———
03Qwen3 VL 4B (Reasoning)+1 altAlibaba55.2VENDOR———
04MiMo Embodied 7BXiaomi48.5VENDOR———
05Qwen3 VL 2BAlibaba48VENDOR———
06Kimi VL A3B InstructMoonshot37.4———
07GPT-4oOpenAI34$5.0/M$15.0/M$0.294
08Gemma 3 12BGoogle32.4$0.05/M$0.15/M$0.003
09Phi 4 R V 15BMicrosoft25.5———
10Phi 4 Multimodal InstructMicrosoft24.09———
◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.

Verification: 10 models scored · 0 independently verified · 5 vendor cross-reference · 5 vendor-reported · 0 with source disagreement. How these tiers are assigned