Which model leads OmniDocBench 1.5 (average edit distance, lower is better)?

Q: Which model leads OmniDocBench 1.5 (average edit distance, lower is better)?

Gemma 3 27B tops the board at 0.365

OmniDocBench 1.5 (average edit distance, lower is better) Leaderboard — Model Scores

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

The field · 6 measured

0.1310.365

6 models measured, most on the official card harness. Gemma 3 27B tops the board at 0.365.

6 measured·5 new this week·Lifecycle Updated Oct 8, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 6Bars show 0.131–0.365 · ◆ solid = first-party or better · outlined = arm's-length

01Gemma 3 27BNEW+27 altsGoogle0.365$0.08/M$0.16/M$0.329

02Gemma 4 E2BNEW+36 altsGoogle0.29———

03Gemma 4 E4BNEW+30 altsGoogle0.181$0.02/M$0.10/M$0.331

04Gemma 4 12B+22 altsGoogle0.164$0.10/M$0.30/M$1.220

05Gemma 4 26B A4BNEW+30 altsGoogle0.149$0.07/M$0.34/M$1.376

06Gemma 4 31BNEW+26 altsGoogle0.131$0.17/M$0.40/M$2.156

◆ 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: 6 models scored · 0 independently verified · 6 vendor-reported · 0 with source disagreement. How these tiers are assigned