Domains: MathCodingScienceBusiness
Skills: ReasoningCodingInstruction FollowingAgenticWriting

Four-Factor Integrity Breakdown & Primary-Source Evidence

Privacy

Weight 40%
7.5/10
Access Model: Hybrid Public / Private
"The current LiveBench release is 2025-04-25; however, not all questions for this release are public on Huggingface. In order to evaluate all categories, you will need to pass --livebench-release-option 2024-11-25 to all scripts to use the most recent public questions."
Primary source (GitHub) ↗ Verified 2026-11-17

Originality

Weight 25%
7.2/10
Reuse Profile: Low Reuse (<25%)
"LiveBench is designed to limit potential contamination by releasing new questions monthly, as well as having questions based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses."
Primary source (GitHub) ↗ Verified 2026-11-17

Freshness

Weight 20%
10/10
Initial Release: 2024-06Latest Refresh: 2026-06Rolling Updates: Yes
"Adds the LiveBench-2026-06-25 release (agentic coding v2) to the old leaderboard site — a new release date, refreshed agentic coding data, and one new model."
Primary source (GitHub) ↗ Verified 2026-11-17

Grading

Weight 15%
10/10
Integrity: Independent Offline (Leak-Verified)Sandbox: offline isolated
"# ANSWER-LEAK / NETWORK ISOLATION: run the agent container with no network."
Primary source (GitHub) ↗ Verified 2026-11-17

Frontier Model Standings on LiveBench

Top 15 active models tracked
Rank Model Official Score Percentile
#1Gemini 4 Neon xhigh Google88.4%100.0p
#2GPT-6.2 Astra max OpenAI86.5%100.0p
#3GPT-6 Astra max OpenAI85.7%100.0p
#4Claude Fable 5.2 max Anthropic85.6%97.9p
#5Claude Fable 5.1 max Anthropic85.1%94.5p
#6Claude Opus 5.5 max Anthropic84.7%91.2p
#7Gemini 4 Carbon high Google84.7%87.8p
#8GPT-6.1 Sol max OpenAI83.6%84.5p
#9GPT-6.2 Sol max OpenAI82.5%81.1p
#10GPT-6.1 Terra max OpenAI81.1%77.8p
#11Gemini 4 Argon high Google80.5%74.4p
#12Gemini 4 Hydrogen xhigh Google79.0%71.1p
#13Claude Sonnet 5.5 max Anthropic78.5%67.7p
#14Qwen4 Max xhigh Alibaba76.6%64.3p
#15DeepSeek V4.1 Flash 1017 max DeepSeek73.3%60.9p

Embeddable Audit Badge

View Markdown (.md)

Benchmark maintainers can embed this live SVG badge in GitHub READMEs or documentation:

BenchAudit LiveBench Tier A 8.3/10
[![BenchAudit Tier A](https://benchaudit.org/badges/livebench.svg)](https://benchaudit.org/benchmarks/livebench/)