Seed-OSS-36B-Instruct — benchmark results

ByteDance's open Seed-OSS 36B instruct model (512K context). Provider: ByteDance. Released 2025-10-01. Access: Open.

Unified ELO 1610 ± 17, rank #455 of 1776 rated models, from 79 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
SuiChat-CN74.5Standard F1 (self-reported)90.7
AA LiveCodeBench76.51Pass@1 (%)87.4
AA AIME 202584.67Accuracy (%)81.9
AA MMLU-Pro81.54Accuracy (%)75.6
AA-LCR57.7Score (self-reported)72.4
AI Chess Leaderboard (Reasoning)818Elo70.6
AA Long Context Reasoning57.67Accuracy (%)67.8
AA Omniscience - Software Engineering (SWE) - Go24Accuracy (%)66.1
UGI - Natural Intelligence25.9NatInt Score65.2
AA SciCode36.46Accuracy (%)59.5
AA Omniscience - Software Engineering (SWE) - HTML34Accuracy (%)59
AA Humanity's Last Exam9.13Accuracy (%)58.6

Interactive version: theaggregate.ai/model?slug=seed-oss-36b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.