--- id: "/evaluation-benchmarks/safety-efficiency-evals/" title: "safety & efficiency evals: 44 curated resources" url: "https://skillissue.dev/evaluation-benchmarks/safety-efficiency-evals/" --- # safety & efficiency evals: 44 curated resources safety & efficiency evals [Beginner](/evaluation-benchmarks/beginner/index.md) [Intermediate](/evaluation-benchmarks/intermediate/index.md) [Advanced](/evaluation-benchmarks/advanced/index.md) [Expert](/evaluation-benchmarks/expert/index.md) 44 resources · page 1 of 2 - ### [AILuminate](/evaluation/benchmarks-and-leaderboards/ailuminate/index.md) by [MLCommons](https://mlcommons.org/) MLCommons AILuminate benchmark that rates the safety of general-purpose chat models' responses to hazardous text prompts. entity benchmark International en - ### [AIR-Bench 2024](/evaluation/benchmarks-and-leaderboards/air-bench-2024/index.md) by [Stanford Center for Research on Foundation Models](https://crfm.stanford.edu/) AIR-Bench 2024 GitHub repo from Stanford CRFM: safety benchmark with a risk taxonomy drawn from regulations and company policies. entity benchmark US (Stanford CRFM) en - ### [CAICT 2026 launch notice (WeChat)](/evaluation/benchmarks-and-leaderboards/caict-2026-launch-notice-wechat/index.md) by [China Academy of Information and Communications Technology (CAICT)](https://www.caict.ac.cn/) CAICT WeChat notice in Chinese announcing its 2026 AI evaluation rounds. entity news China [STATE] zh - ### [Canadian Artificial Intelligence Safety Institute (CAISI)](/evaluation/benchmarks-and-leaderboards/canada-aisi-caisi/index.md) by [Government of Canada (ISED)](https://ised-isde.canada.ca/) Canadian Artificial Intelligence Safety Institute page on the ISED government site. entity org Canada en|fr - ### [ChinAI #261: First results from CAICT's AI Safety Benchmark](/evaluation/benchmarks-and-leaderboards/chinai-261-first-caict-ai-safety-benchmark-results/index.md) by [Jeffrey Ding (ChinAI)](https://chinai.substack.com/) ChinAI newsletter issue translating the first results of CAICT's AI Safety Benchmark for Chinese language models. entity benchmark China [STATE] subject en - ### [CValues](/evaluation/benchmarks-and-leaderboards/cvalues/index.md) by [X-PLUG](https://github.com/X-PLUG) CValues GitHub repo: benchmark and research on evaluating the values and safety of Chinese large language models. entity benchmark China zh - ### [CyberSecEval 4](/evaluation/benchmarks-and-leaderboards/cyberseceval/index.md) by [Meta Llama](https://llama.meta.com/) Meta CyberSecEval 4 documentation: open benchmarks for cybersecurity risks and capabilities of LLMs. entity repo US en - ### [FAR.AI AI Security Leaderboard](/evaluation/benchmarks-and-leaderboards/far-ai-ai-security-leaderboard/index.md) by [FAR.AI](https://far.ai/) FAR.AI AI Security Leaderboard comparing model safeguards against a taxonomy of attacks with a published methodology. entity benchmark US / International en - ### [AI Safety Index: Summer 2026 (Future of Life Institute)](/evaluation/benchmarks-and-leaderboards/fli-ai-safety-index-summer-2026/index.md) by [Future of Life Institute](https://futureoflife.org/) Future of Life Institute AI Safety Index Summer 2026: expert grades of AI companies across safety and security domains. entity leaderboard US (non-profit) en - ### [ForesightSafety Bench](/evaluation/benchmarks-and-leaderboards/foresightsafety-bench/index.md) by [Haibo Tong et al.](https://openalex.org/authors/A5123692526) ForesightSafety Bench paper: a frontier-risk evaluation framework covering agentic autonomy, AI-for-science, embodied and catastrophic risks. entity benchmark China (per ChinAI search; affiliations not shown on abs page) en - ### [HallusionBench](/evaluation/benchmarks-and-leaderboards/hallusionbench/index.md) by [Tianyi Lab](https://tianyizhou.github.io/) HallusionBench GitHub repo (CVPR 2024): benchmark of language hallucination and visual illusion in VLMs. entity repo US / International en - ### [HarmBench](/evaluation/benchmarks-and-leaderboards/harmbench/index.md) by [Center for AI Safety (HarmBench)](https://www.harmbench.org/) HarmBench site: standardised evaluation framework for automated red-teaming attacks and model refusal defences. entity benchmark US / International en - ### [IPA press release: AI safety evaluation demonstration and environment report meeting](/evaluation/benchmarks-and-leaderboards/ipa-press-release-aisi-tf-meeting-2026-10-09/index.md) by [Information-technology Promotion Agency, Japan (IPA)](https://www.ipa.go.jp/) IPA press release in Japanese on a briefing about AI safety evaluation trials and its open evaluation environment. entity news Japan ja - ### [JailbreakBench](/evaluation/benchmarks-and-leaderboards/jailbreakbench/index.md) by [JailbreakBench](https://github.com/JailbreakBench) JailbreakBench site: benchmark and leaderboard tracking jailbreak attacks and defences on LLMs with a fixed set of behaviours. entity benchmark International en - ### [Japan AI Safety Institute](/evaluation/benchmarks-and-leaderboards/japan-ai-safety-institute/index.md) by [Japan AI Safety Institute](https://aisi.go.jp/) Japan AI Safety Institute: government body developing AI safety evaluation methods, guides and standards. entity org Japan ja - ### [米OpenAI社とMOCを締結しました (Japan AISI)](/evaluation/benchmarks-and-leaderboards/japan-aisi-openai-moc-2026-05-29/index.md) by [Japan AI Safety Institute](https://aisi.go.jp/) Japan AISI announcement (May 2026): memorandum of cooperation signed with OpenAI. entity standard-policy Japan ja - ### [Korea AISI](/evaluation/benchmarks-and-leaderboards/korea-aisi/index.md) by [Korea AISI](https://aisi.re.kr/) Korea AI Safety Institute: government institute for AI safety evaluation, research and international cooperation. entity org Korea ko - ### [MLCommons AILuminate](/evaluation/benchmarks-and-leaderboards/mlcommons-ailuminate/index.md) by [MLCommons](https://mlcommons.org/) MLCommons AILuminate page: safety benchmark for general-purpose chatbots, with jailbreak and agentic tracks listed. entity benchmark US (industry consortium) en|fr - ### [MLPerf Client](/evaluation/benchmarks-and-leaderboards/mlperf-client/index.md) by [MLCommons](https://mlcommons.org/) MLCommons MLPerf Client page: benchmark for LLM and AI workload performance on personal computers. entity benchmark International en - ### [MLPerf Inference](/evaluation/benchmarks-and-leaderboards/mlperf-inference/index.md) by [MLCommons](https://mlcommons.org/) MLCommons MLPerf Inference Datacenter results: standardised throughput and latency of AI systems across models. entity leaderboard West en - ### [MLPerf Tiny](/evaluation/benchmarks-and-leaderboards/mlperf-tiny/index.md) by [MLCommons](https://mlcommons.org/) MLCommons MLPerf Tiny page: inference benchmark for embedded and low-power devices, with published results. entity benchmark International en - ### [MLPerf Training](/evaluation/benchmarks-and-leaderboards/mlperf-training/index.md) by [MLCommons](https://mlcommons.org/) MLCommons MLPerf Training page: benchmark measuring time to train models to target quality, with published results. entity benchmark International en - ### [NIST CAISI](/evaluation/benchmarks-and-leaderboards/nist-caisi/index.md) by [NIST](https://nist.gov/) NIST CAISI: US government center evaluating AI models and developing measurement standards. entity leaderboard West en - ### [Phare](/evaluation/benchmarks-and-leaderboards/phare/index.md) by [Giskard](https://www.giskard.ai/) Giskard Phare site: multilingual benchmark of LLM safety and security, including hallucination, bias and harmful content. entity benchmark France / International mul - ### [S-Eval](/evaluation/benchmarks-and-leaderboards/s-eval/index.md) by [IS2Lab](https://is2lab.github.io) S-Eval GitHub repo: about 220,000 safety prompts, including adversarial variants, for automated LLM safety evaluation. entity benchmark China (Zhejiang Univ. + Alibaba AAIG, inferred) zh|en - ### [S-Eval leaderboard (Alibaba)](/evaluation/benchmarks-and-leaderboards/s-eval-leaderboard-alibaba/index.md) by [Alibaba (S-Eval)](https://www.alibabagroup.com/) Alibaba S-Eval leaderboard in Chinese: safety evaluation rankings of large language models. entity leaderboard China (commercial) zh - ### [SafetyBench](/evaluation/benchmarks-and-leaderboards/safetybench/index.md) by [thu-coai](https://coai.cs.tsinghua.edu.cn/) SafetyBench GitHub repo from Tsinghua: multiple-choice safety benchmark for LLMs across seven categories, in Chinese and English. entity benchmark China zh - ### [SALAD-Bench](/evaluation/benchmarks-and-leaderboards/salad-bench/index.md) by [OpenSafetyLab](https://github.com/OpenSafetyLab) SALAD-Bench GitHub repo: hierarchical safety benchmark of harmful questions plus the MD-Judge evaluator model. entity benchmark China [STATE] (Shanghai AI Lab per memory; page names no institution) en - ### [Singapore AISI (Digital Trust Centre)](/evaluation/benchmarks-and-leaderboards/singapore-aisi-digital-trust-centre/index.md) by [Singapore AI Safety Institute](https://www.sgaisi.sg/) Singapore AI Safety Institute (Digital Trust Centre): research on AI safety science and governance. entity org Singapore en - ### [StrongREJECT](/evaluation/benchmarks-and-leaderboards/strongreject/index.md) by [StrongREJECT team (UC Berkeley)](https://github.com/dsbowen/strong_reject) StrongREJECT documentation: benchmark package that scores whether jailbreaks yield useful harmful outputs rather than mere compliance. entity benchmark US / International en - ### [SuperCLUE-Safety](/evaluation/benchmarks-and-leaderboards/superclue-safety/index.md) by [CLUE benchmark](https://www.CLUEbenchmarks.com) SuperCLUE-Safety GitHub repo: a multi-turn adversarial safety benchmark for Chinese large language models. entity benchmark China zh - ### [UK AI Security Institute](/evaluation/benchmarks-and-leaderboards/uk-ai-security-institute/index.md) by [UK AI Security Institute](https://www.aisi.gov.uk/) UK AI Security Institute: government body publishing frontier AI evaluations and the Inspect evaluation framework. entity leaderboard West en [Next page](/evaluation-benchmarks/safety-efficiency-evals/page/2/index.md) [Plain-text Markdown copy](/evaluation-benchmarks/safety-efficiency-evals/index.md)