List your company

Instruction Following on MultiChallenge

69.3ScoreGemini-3-Flash
36.467244.991153.51562.0389Jan 14, 2026Feb 10, 2026Mar 10, 2026Apr 7, 2026May 5, 2026Jun 2, 2026Jun 30, 2026

Evaluation results

MethodLinks
Gemini-3-FlashTier=High2026.0669.3—
Gemini-3-pro High2026.0668.7—
Seed2.0 Pro2026.0668.3—
ERNIE 5.02026.0265.98—
DS-v4-ProParameter Count=1.6T-A49B2026.0664.1—
Qwen-3.5Parameter Count=397B-17B2026.0663.9—
Nemotron 3 UltraParameter Count=550B-A55B2026.0663.8—
DS-v4-FlashParameter Count=284B-A13B2026.0663.5—
Seed2.0 Lite2026.0663.2—
Kimi-K2.6Parameter Count=1T-A32B2026.0663.1—
GLM-5.1Parameter Count=744B-A40B2026.0663—
STEP3-VL-10BNumber of Parameters=10B2026.0162.64—
Gemini 3-Pro2026.0262.5—
Seed2.0 Mini2026.0661.1—
GPT-5.2 High2026.0659.5—
GPT-5-miniTier=High2026.0659—
Claude-Opus-4.52026.0659—
GPT-5 (High)2026.0258.3—
Claude-Sonnet-4.52026.0657.3—
Gemini 2.5-Pro2026.0251.5—
Qwen3-VL ThinkingNumber of Parameters=8B2026.0149.82—
MiMo-VL RL-2508Number of Parameters=7B2026.0144.69—
MiniMax-2.7Parameter Count=230B-A10B2026.0642.5—
GLM-4.6V FlashNumber of Parameters=9B2026.0142.49—
DS V3.2-Thinking2026.0242.43—
InternVL 3.5Number of Parameters=8B2026.0137.73—
AutoIF-RLVR2026.06—45.4
QUBRIC2026.06—44.5
RM-RLHFlength_penalty=none2026.06—45.7
RM-RLHFlength_penalty=1.02026.06—41.8
Rubric-RLquery=Original Query2026.06—44.5
SFT Baseline2026.06—40.2