Which model leads OCRBench v1?

Qwen3.5-2B leads the OCRBench v1 board of 7 measured models on the Vector Wire tracker at 84.4. Every score is source-attributed with a verification tier.

OCRBench v1 Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensofficial card harness

The field · 7 measured

70.284.4

7 models measured, most on the official card harness. Qwen3.5-2B tops the board at 84.4.

7 measured·5 new this week·Lifecycle Updated Sep 12, 2026Basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 7Bars show 70.2–84.4· ◆ solid = first-party or better · outlined = arm's-length

01Qwen3.5-2BNEWAlibaba84.4———

02LFM2.5-VL-3BLiquid AI84.2———

03InternVL 3.5 2BNEWOpenGVLab83.9———

04InternVL 3.5 4BNEWOpenGVLab82———

05LFM2-VL-3BLiquid AI81.7———

06gemma-4-E4B-itNEWGoogle73.5$0.02/M$0.10/M$0.001

07gemma-4-E2B-itNEWGoogle70.2———

◆ 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.