Doubao-Seed-1-8-251228: benchmark results

Provider: ByteDance. Released 2025-12-28. Access: API.

Aggregate Rating 1712 ± 16, rank #282 of 10800 rated models, from 46 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
ChronoPhyBench - Frame Sorting (4 Frames)46.5Accuracy (%) on ChronoPhyBench chronological sorting of 4 he100
ChronoPhyBench - Next-State Selection (6 Options)53Accuracy (%) on ChronoPhyBench next-state frame selection wi87.5
CDH-Bench (Binary QA)76.67Counterfactual accuracy (%) on the 300 counterfactual images86.7
ChronoPhyBench - Next-State Selection (3 Options)55Accuracy (%) on ChronoPhyBench next-state frame selection wi85.7
ChronoPhyBench - Frame Sorting (3 Frames)61Accuracy (%) on ChronoPhyBench chronological sorting of 3 he85
BABE (Biology Arena)43.27Average score (%), the paper's fixed weighting of its strong81.2
BABE (Biology Arena) - Strong Correlation43.38Score (%) on the strong-correlation triplets, where each que81.2
BABE (Biology Arena) - Weak Correlation43.16Score (%) on the weak-correlation triplets of independent qu81.2
CDH-Bench (Binary QA) - Relational Anomalies80Counterfactual accuracy (%) on the 100 relational-anomaly im78.6
OpenCompass Language - Creation (CompassBench 2601)91Score (%)78
SuperCLUE General (2025 Annual) - Precise Instruction Following32.6Score75.8
CDH-Bench (Multiple-Choice QA) - Counting Anomalies71Counterfactual accuracy (%) on the 100 counting-anomaly imag71.4

Interactive version: theaggregate.ai/model?slug=doubao-seed-1-8-251228 · How It Works · Data refreshed daily, snapshot 2026-11-13.