MAI Code 2: Microsoft's Cost Claims Need Independent Testing
Microsoft's headline math is striking: MAI Code 2 reaches 98% of GPT-6.1 Sol's performance at 40% of the cost, and 96% of Claude Opus 5.5's at 5%. The...

Created by weiqun zou
Latest news, benchmarks, and use‑case guides for AI coding assistants and autonomous agents
Explore the latest content tracked by AI Coding Tools Digest
Microsoft's headline math is striking: MAI Code 2 reaches 98% of GPT-6.1 Sol's performance at 40% of the cost, and 96% of Claude Opus 5.5's at 5%. The...
As AI-generated code grows, the risks of untested vibe coding are becoming harder to ignore:
This roundup tracked 18 releases in 24 hours. The coding-tool updates that matter most:
The paper reports 91.4% on LoCoMo and 72.6% on LongMemEval-S, leading four of five QA benchmarks.
The Theia AI guide shows how an extensible IDE can host tailored assistants:
The piece treats the shift from autocomplete to agents as a control problem, not a question of output volume.
Two new tools wrap coding agents in delivery infrastructure rather than just generating code. Tencent's LoopForge doesn't replace tools like Claude...
A device-code request approved on GitHub's real site can still hand access to someone else. In a controlled Copilot CLI simulation, researchers showed...
Pangolin describes managing pipeline network access as code: rules are written in YAML, changes are reviewed in pull requests, and CI applies them. If...
A benchmark reports that 64% of inference spend went to runs that didn't pass, across 27,280 runs and 83B tokens. The page is titled "Introducing...
The report describes a 320-billion-parameter mixture-of-experts model with dedicated encoders for molecular graphs and biological sequences, feeding a...
Start by separating the two billing camps. The GLM Coding Plan and Kimi Code are quota-based subscriptions, while the DeepSeek API bills per token....
Designing around AI shifts the focus from individual coding speed to how work moves between stages. Four changes stand out:
The post frames SWE-CoWork as a test of whether coding agents can be dependable digital engineers on long-horizon tasks. To pass, agents must gather...
Permission layers usually see only the path text an agent requests, but the kernel follows symlinks until it reaches the actual file. A committed...
The Jev MCP server is a Model Context Protocol implementation that links assistants like Claude and Cursor to your Jev account, giving them structured...
Sessionpipe, announced as open source by @micmakarov, says it connects coding-agent sessions "on any harness" and lets developers save, search,...
Kiro's spec-driven flow turns one prompt into requirements, design, and tasks you review before any code is written. The article argues that catching...
The core dispute is whether agent-run checks justify their token cost. One commenter on the Claude Code team's verification post calls some practices...
Two very different threats are emerging. SPILLWAY, a new arXiv preprint, targets hidden instructions in web pages or documents that could redirect...