List your company

Text-based safety moderation on BeaverTails

87.3F1 ScoreRoblox Guard 1.0
28.74843.94959.1574.351Dec 2, 2025Jan 4, 2026Feb 7, 2026Mar 13, 2026Apr 15, 2026May 19, 2026Jun 22, 2026

Evaluation results

MethodLinks
Roblox Guard 1.0Parameter Count=8B2025.1287.3—
Qwen3Guard-8B-strictAccess=Open-Source, Configuration=strict2026.0686.69—
Qwen3Guard-4B-Gen-strictStreaming=no, Moderation Policy=Gen-strict, Backbone Scale=4B2026.0486.6—
Qwen3Guard-8B-Gen-strictStreaming=no, Moderation Policy=Gen-strict, Backbone Scale=8B2026.0486.6—
Llama3-StreamGuard-8BStreaming=yes, Backbone Scale=8B2026.0486.6—
BingoGuardParameter Count=8B2025.1286.4—
Llama3-StreamGuard-3BStreaming=yes, Backbone Scale=3B2026.0486.4—
Llama3-StreamGuard-1BStreaming=yes, Backbone Scale=1B2026.0486.2—
SingGuard-2BAccess=Open-Source2026.0686.01—
Qwen3Guard-Stream-4B-strictStreaming=yes, Moderation Policy=strict, Backbone Scale=4B2026.0486—
Qwen3Guard-Stream-8B-strictStreaming=yes, Moderation Policy=strict, Backbone Scale=8B2026.0485.9—
SingGuard-8BAccess=Open-Source2026.0685.85—
SingGuard-4BAccess=Open-Source2026.0685.82—
YuFeng-XGuard-Reason-8BAccess=Open-Source2026.0685.71—
Qwen3Guard-Stream-8B-looseStreaming=yes, Moderation Policy=loose, Backbone Scale=8B2026.0485.5—
Qwen3Guard-Stream-4B-looseStreaming=yes, Moderation Policy=loose, Backbone Scale=4B2026.0485.2—
Qwen3Guard-8B-looseAccess=Open-Source, Configuration=loose2026.0684.93—
ShieldGemmaParameter Count=7B2025.1284.8—
Qwen3Guard-Stream-0.6B-strictStreaming=yes, Moderation Policy=strict, Backbone Scale=0.6B2026.0484.5—
GuardReasoner-VL-7BAccess=Open-Source2026.0684.49—
WildGuardParameter Count=7B2025.1284.4—
MD-JudgeSize=7B2026.0384—
Qwen3Guard-Stream-0.6B-looseStreaming=yes, Moderation Policy=loose, Backbone Scale=0.6B2026.0484—
OMNIGUARD-7BSize=7B2025.1283.980.5
GPT-4o2025.1283.8—
Qwen2.5-72BSize=72B2025.1283.680.5
GPT-4oSize=-2025.1283.572.6
WildguardSize=7B2026.0383—
FanarGuardSize=4B2026.0383—
ThinkGuardSize=8B2025.1282.781.6
Qwen3-VL-235BAccess=Open-Source2026.0682.15—
FanarGuardSize=4B2026.0382—
Qwen3-235BSize=235B2025.1281.977.1
OMNIGUARD-3BSize=3B2025.1281.882.6
GPT-5.1Access=Closed-Source2026.0680.42—
PolyGuard-MinistralSize=8B2026.0380—
PolyGuard-QwenSize=7B2026.0380—
Gemini3-ProAccess=Closed-Source2026.0679.6—
PolyGuard-MinistralSize=8B2026.0379—
PolyGuard-QwenSize=7B2026.0378—
WildGuardAccess=Open-Source2026.0677.89—
NemoGuardParameter Count=8B2025.1277.6—
LLaMA-3.3-70BSize=70B2025.1276.871.3
ShieldGemma-2bSize=2B2026.0376—
Qwen2.5-7BSize=7B2025.1275.372.6
GraniteGuardianAccess=Open-Source2026.0672.91—
LLaMA Guard 2Size=8B2025.1272.373.5
LLaMA Guard 3Size=8B2025.1271.273.5
PolyGuard-SmolSize=0.5B2026.0371—
PolyGuard-SmolSize=0.5B2026.0371—
ShieldGemma-2bSize=2B2026.0371—
Llama-Guard-3Size=8B2026.0370—
LlamaGuard3Parameter Count=8B2025.1269.7—
ShieldGemma-9BAccess=Open-Source2026.0669.62—
Llama Guard 3Access=Open-Source2026.0667.78—
Llama-Guard-3Size=8B2026.0366—
Qwen2.5-Omni-7BSize=7B2025.1258.455.4
WildguardSize=7B2026.0348—
LLaMA Guard 1Size=7B2025.1238.155.9
MD-JudgeSize=7B2026.0331—