1
I benched frontier AI LLM models on politics, ethics, and personality traits
前沿AI模型的政治立场、伦理取向与人格特质，一次基准测试看透底牌
Article URL: https://www.blackbench.ai/ Comments URL: https://news.ycombinator.com/item?id=49382540 Points: 3 # Comments: 2
2
SpaceXAI debuts Grok 4.6, overtaking Kimi K3's performance and matching GPT-5.6 Sol for world's third best on Artificial Analysis
Grok 4.6正式亮相，性能超越Kimi K3、紧咬GPT-5.6 Sol，还以不到半价的API成本抢占市场。AI模型混战再升级，值得关注。
Elon Musk's company SpaceXAI, formerly known as xAI, has released Grok 4.6 , its latest frontier AI model, with a focus on long-running agents, c…
3
Show HN: Benchmarklist: track AI benchmarks (2.4k+), models, and capabilities
一键追踪2400+AI基准测试，快速查看模型能力排名与评分。
We’re launching BenchmarkList: one place to track AI benchmarks, models, and capabilities. It’s surprisingly hard to get a complete picture of what AI…
4
Separating signal from noise in coding evaluations
OpenAI揭秘热门编码基准测试的缺陷，重新审视AI模型评估的可靠性
A new analysis from OpenAI reveals issues in SWE-Bench Pro, a popular coding benchmark, raising concerns about reliability and accuracy in evaluating …
5
BenchX: Benchmarking AI Models for Cancer Detection and Localization with Demographic and Protocol Biases
聚焦AI癌症检测模型的基准测试，揭露人口统计与扫描协议偏差如何影响诊断性能，为医疗AI公平性提供关键参考。
arXiv:2606.24883v1 Announce Type: new Abstract: Artificial intelligence (AI) has achieved remarkable success in medical imaging, but it is widely reco…
6
A Reproducible Semantic Benchmark for Multivendor DSM-to-CLI Translation
网络设备配置翻译迎来可复现基准，多厂商DSM转CLI的语义评测新标准，用AI打通状态模型与命令行鸿沟。
arXiv:2606.20564v1 Announce Type: cross Abstract: Translating high-level network intents into correct multivendor configurations remains a central cha…
7
WebSP-Eval: Evaluating Web Agents on Website Security and Privacy Tasks
首个评估AI网页智能体在网站安全与隐私任务上的基准，揭示现有模型的脆弱性
arXiv:2604.06367v2 Announce Type: replace-cross Abstract: Web agents automate browser tasks, ranging from simple form completion to complex workflows …
8
SupraBench: A Benchmark for Supramolecular Chemistry
首个超分子化学AI基准测试，推动分子计算与新药研发
arXiv:2606.13477v1 Announce Type: cross Abstract: Supramolecular chemistry, which includes the study of non-covalent host-guest assemblies, has advanc…
9
What AI benchmarks miss about real-world performance
AI基准测试忽略的致命短板：买够GPU却卡在数据路径的突发瓶颈上。
Presented by F5 Enterprise AI teams have spent years solving for compute, securing GPU allocations, negotiating cloud capacity, and benchmarking train…
10
Can AI Reason Like an Urban Planner? Benchmarking Large Language Models Against Professional Judgment
大型语言模型能否像城市规划师一样推理？研究用专业判断基准测试AI，揭示AI与人类专家的能力边界。
arXiv:2606.11678v1 Announce Type: new Abstract: Problem, Research Strategy, and Findings: The rise of large language models (LLMs) raises a key questi…
11
These LLMs are the best at resisting Russian propaganda
最新基准测试显示，Anthropic的Claude模型在抵制俄罗斯宣传方面表现最优，Opus 4.7获得“杰出”评级。
Estonian government benchmark shows how dozens of models combat Russia's "strategic narratives."
12
PRAIB: Peer Review AI Benchmark of Behaviour of LLM-Assisted Reviewing
首个专门评估LLM辅助同行评审行为的基准，揭示AI评审的可靠性边界。
arXiv:2605.29815v1 Announce Type: new Abstract: The growing number of submitted papers has motivated the exploration of Large Language Models (LLMs) a…
13
AI Cartography: Mapping the Latent Landscape of AI Benchmark Ecosystems
揭示AI排行榜得分背后的测量噪声，并提出框架量化潜在能力差异，颠覆你对基准测试的认知。
arXiv:2605.25272v1 Announce Type: new Abstract: While aggregate leaderboard scores drive AI development, they contain substantial measurement noise wh…
14
The Evaluation Trap: Benchmark Design as Theoretical Commitment
AI基准测试暗藏理论假设，窄化进步定义，警惕评估陷阱重塑能力概念
arXiv:2605.14167v1 Announce Type: new Abstract: Every AI benchmark operationalizes theoretical assumptions about the capability it claims to assess. W…
15
IPR-1: Interactive Physical Reasoner
看AI如何像人类一样通过观察和交互学习物理与因果，1000+异构游戏基准测试揭秘突破！
arXiv:2511.15407v3 Announce Type: replace Abstract: Humans learn by observing, interacting with environments, and internalizing physics and causality.…