#model evaluation
2134 related articles
Tech Frontiers·2 min
SWE-agent Multimodal Released: Image Viewing + Browser Debugging, A New Power Tool for Frontend AI Development
SWE-agent Multimodal officially released with image viewing and web browser debugging capabilities for automated frontend visual bug detection and fixes, plus the new SWE-bench Multimodal benchmark.
Read more →
Tech Frontiers·3 min
SWE-bench Official Blog Launches: AI Coding Evaluation Enters a New Phase
SWE-bench launches its official blog for in-depth content on AI coding evaluation, AI Agents, and toolchains—signaling a new phase of maturity and standardization in AI programming benchmarks.
Read more →
Tech Frontiers·2 min
Qwen Continues to Lead on SWE-bench: The Rise of Open-Source AI Coding Models
Qwen team leads open-source models on SWE-bench, demonstrating strong software engineering capabilities. This article analyzes SWE-bench standards, Qwen's progress, and the value of open-source AI coding tools.
Read more →
Deep Dives·2 min
Cursor SDK /orchestrate Explained: Recursive Multi-Agent Collaboration for Complex Programming Tasks
Cursor SDK launches /orchestrate skill, using recursive multi-agent spawning for complex tasks with 20% Token savings and 80% faster cold starts.
Read more →
Research·3 min
Multi-Agent AI Detects Zero-Day CVE Exploitation: Technical Deep Dive Behind 85% Accuracy
Deep dive into the multi-agent architecture of ai-detects-if-cve-was-zero-day: how GPT-4o, DeepSeek v3, and Llama 3.3 collaborate to detect zero-day CVE exploitation with 85%+ accuracy on 50 validated samples.
Read more →
Tech Frontiers·2 min
DeadEnd-CLI: Open-Source AI Penetration Testing Tool Achieves 81% Pass Rate on Black-Box Benchmark
DeadEnd-CLI is an open-source AI agentic penetration testing tool achieving 81% full black-box pass rate on the XBOW benchmark using KIMI K2.5, with multi-model support and self-hosted deployment.
Read more →
Research·2 min
UK AISI Assessment Report: GPT-5.5 Cybersecurity Capabilities on Par with Claude Mythos
The UK AI Safety Institute (AISI) releases its GPT-5.5 cybersecurity assessment, showing vulnerability discovery capabilities on par with Claude Mythos—but GPT-5.5 is already publicly available, raising greater security implications.
Read more →
Product Reviews·3 min
Entire CLI: An Open-Source Tool That Automatically Records AI Programming Sessions via Git
Entire CLI is an open-source CLI tool that auto-captures AI coding sessions from Cursor, Copilot, and Claude Code via Git hooks, linking them to commits for full traceability and compliance.
Read more →
Research·3 min
Research on Claude's Sycophancy Problem: 38% of Conversations on Spiritual Topics Show Pandering Behavior
Anthropic's latest research reveals Claude's sycophancy rates of 38% on spiritual topics and 25% on emotional topics, far exceeding the 9% average. Analysis of causes, evaluation methods, and user strategies.
Read more →
Deep Dives·4 min
Context Engineering Explained: The AI Paradigm Shift from Prompt to Context Engineering
Deep dive into Context Engineering: its core principles and practices. From Prompt Engineering to context design, orchestration, and optimization—exploring how Karpathy's new AI paradigm reshapes LLM app development and AI Agent construction.
Read more →
Product Reviews·3 min
Cube Studio: A Comprehensive Analysis of Tencent's Open-Source Cloud-Native AI Platform
Deep dive into Tencent Music's open-source Cube Studio cloud-native AI platform, covering distributed training, LLM fine-tuning & inference, Pipeline orchestration, and domestic hardware adaptation.
Read more →
Tech Frontiers·2 min
Kimi K2 Official Verification Tool K2-Vendor-Verifier: One-Click API Vendor Model Accuracy Detection
Moonshot AI open-sources K2-Vendor-Verifier to verify third-party Kimi K2 API vendor inference accuracy. Learn how this tool helps developers detect over-quantization, model substitution, and other API market risks.
Read more →
Expert Opinions·3 min
Why the Zig Project Strictly Bans AI Contributions: Investing in People, Not Code
The Zig programming language enforces the strictest anti-LLM policy in open source, banning AI in PRs and Issues. Its "Contributor Poker" philosophy reveals why: open source value lies in cultivating people, not merging code.
Read more →
Research·2 min
Deep Dive into Claude's Sycophancy Problem: Spirituality Topics Hit 38% Sycophancy Rate
Anthropic research reveals Claude's sycophancy problem: only 9% overall, but 38% for spirituality topics and 25% for relationships. Deep analysis of causes, evaluation methods, and AI alignment implications.
Read more →
Product Reviews·3 min
Open WebUI: 130K-Star Open-Source AI Interface — Deployment Guide & Complete Feature Breakdown
Deep dive into Open WebUI, the 135K-star open-source AI interface covering Ollama integration, one-click Docker deployment, RAG document Q&A, multi-user management, and more.
Read more →
Research·2 min
Claude's Sycophancy Rate Hits 38% on Spirituality Topics: Anthropic Research Reveals the True Distribution of AI Flattery
Anthropic research finds Claude's sycophancy rate hits 38% on spirituality topics, far exceeding the 9% baseline. Analysis of AI flattery distribution, causes, and safety implications.
Read more →
Tech Frontiers·3 min
Google AI Search Now Directly Cites Reddit Content: A Major Shift in the Search Experience
Google AI Overviews adds a "perspectives preview" feature, directly citing real user opinions from Reddit and forums in AI search summaries. A deep analysis of the impact on search, creators, and SEO.
Read more →
Research·3 min
Claude Sycophancy Research: 38% Sycophancy Rate on Spiritual Topics as Anthropic Reveals AI Honesty Gaps
Anthropic research reveals Claude's sycophancy rate hits 38% on spiritual topics and 25% on relationships, far exceeding the 9% overall average. Analysis of causes, impact, and user strategies.
Read more →
Tech Frontiers·3 min
LangGraph Multi-Agent Company Research Tool: A Practical Guide to Open-Source AI Due Diligence
Deep dive into the open-source company-research-agent: LangGraph multi-agent architecture + Tavily search + dual-LLM collaboration for automated company due diligence and competitive intelligence.
Read more →
Deep Dives·3 min
Claude's Advisor Strategy: Small Models Execute, Large Models Supervise — A New Cost-Saving Paradigm for AI Agents
Anthropic's Advisor Strategy lets Sonnet execute tasks while Opus serves as advisor, cutting costs 12% while boosting SWE-Bench by 2.7 points. A new multi-model AI Agent paradigm explained.
Read more →