Seed 2.0 Pro — benchmark results

ByteDance's flagship Seed 2.0 Pro reasoning and agent model. Provider: ByteDance. Released 2026-02-16. Access: API.

Unified ELO 1687 ± 24, rank #299 of 1776 rated models, from 36 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
VitaBench 2.047.4Avg@4 Full Context (self-reported)94.7
MMGist58.2Macro ↑ (self-reported)92.3
Position Bias (Lechmazur)28Order Flip % (lower is better)91.4
ZeroEval GPQA Diamond88.9GPQA Diamond Score88.9
CC-OCR V272.15Average (self-reported)85.7
Persuasion (Lechmazur)1.64Average Persuasion Strength85.7
ClawProBench61.07Final Score (self-reported)83.9
OpenClawProBench68.3Overall Score (%)78.8
VisualNeedle36.8w/ Tools Acc. (%) (self-reported)75
LiveSecBench63.04Overall Score (%)71.4
LLM Stats (AIME 2026)94.2Score (%)68.8
Visual Aesthetic Benchmark (VAB)51.3Overall Top-1 ap@1 (self-reported)66.7

Interactive version: theaggregate.ai/model?slug=seed-2-0-pro · How the rankings work · Data refreshed daily, snapshot 2026-07-22.