Kimi K2.7-Code cuts thinking tokens 30% — but practitioners say the benchmarks don't check out

Kimi K2.7-Code claims 30% fewer thinking tokens and a drop-in API swap path, but independent benchmarks show kernel ...

AI Coding Agents Write 180% More Code But Ship Only 30% More Software

AI coding agents boost code output by 180% but shipping rises only 30%, MIT finds. Why private data access beats benchmark ...
Morning Overview on MSN

Microsoft’s new MAI-Code tool turns plain-English descriptions into working app code

Microsoft has introduced MAI-Code, a tool designed to convert plain-English descriptions into functional application code.

Gomboc AI Publishes First Open Benchmark for AI Code Remediation

15 cloud scenarios. 43 merge-ready fixes. 100% loop closure. 12 minutes and $17 to author once; seconds and zero-cost ...
Morning Overview on MSN

The newest Anthropic model just took the top spot on the Super-Agent benchmark — the only AI to finish every test case end-to-end and beat OpenAI’s GPT-5.5

Anthropic’s latest AI model has reportedly reached the top of the Super-Agent benchmark, a grueling test of whether an AI ...
NewsBytes

Xiaomi says its new AI coding model beats Claude Code

Xiaomi has launched Mimo Code v0.1.0, an open-source AI coding tool that reportedly outperforms Anthropic's Claude Code on ...

Xiaomi’s latest AI coding tool claims to outperform Claude Code on complex tasks

The open-source AI coding assistant is designed for long-running software projects and, according to Xiaomi's own benchmarks ...
SD Times

Beyond Benchmarks: Measuring the True Cost of AI-Generated Code

Value stream management involves people in the organization to examine workflows and other processes to ensure they are deriving the maximum value from their efforts while eliminating waste — of ...

DeepSWE blows up the AI coding leaderboard, crowns GPT-5.5, and finds Claude Opus exploiting a benchmark loophole

DeepSWE puts GPT-5.5 atop the AI coding leaderboard while raising new questions about Claude Opus, SWE-Bench Pro, and ...
Crypto Briefing

Claude now authors over 80% of code merged into its own codebase

Anthropic reveals Claude Code now writes over 80% of merged production code, up from low single digits in early 2025, reshaping AI development and engineer ...

Microsoft’s multi-agent AI system tops Anthropic’s Mythos on cybersecurity benchmark

Microsoft's new vulnerability-scanning system, codenamed MDASH, scored 88.45% on the CyberGym benchmark, surpassing single-model systems from Anthropic and OpenAI by using more than 100 specialized AI ...
InfoWorld

Why benchmarks are key to AI progress

Researchers are racing to develop more challenging, interpretable, and fair assessments of AI models that reflect real-world use cases. The stakes are high. Benchmarks are often reduced to leaderboard ...