Topic
AI RESEARCH
All blog posts, tools, and guides about AI Research from Developers Digest.
14 resources - 14 posts
Blog Posts
View in blog →
The ICML 2026 Agent Reproduction Audit: 23% of Examined Papers Had Falsified or Contested Claims
Hugging Face's open challenge used 1,200+ participants and their coding agents to attempt 2,226 ICML 2026 papers claim by claim. 51% had claims independently verified, 23% had a falsified or contested claim, and four documented falsifications include a spotlight theorem that fails after step 224.
7 min read
CLAUDE.md Files Never Stop Growing: A New Paper Names the Mechanism
A study of 247,694 instruction lifetimes in 1,867 repositories shows agentic prompt files grow +226% on average because the reasoning behind each rule decays. Comments encoding that reasoning remove 99.3% of the excess.
7 min read
The $44 Compiler: Persistent Projects Beat Persistent Agents
EvoX Genesis built a 250k-line Rust C compiler with DeepSeek V4 Flash for $44 in tokens by making the project the persistent thing and keeping agents finite-lived. The paper's three runs, the design that made them possible, and what it says about agent memory.
6 min read
OpenAI Publishes Ten Decade-Open Math Proofs, Each Formalized in Lean
OpenAI's next model, codenamed Astra, produced results on ten problems open for at least a decade - including non-sofic groups and Erdős problems 146, 180, and 183 - with every argument formalized as a Lean certificate.
6 min read
AGENTS.md Configuration Smells: 91% of Popular Repos Get One of Six Wrong
A SCAM 2026 study of 100 top-starred repos catalogs six configuration smells in AGENTS.md and CLAUDE.md files: Lint Leakage in 62%, Context Bloat in 42%, Skill Leakage in 35%. Only 9 of 100 files were smell-free.
6 min read
Coding Agents Almost Never Read Open Source Contribution Rules: RepoComplianceBench Study
A new 106-issue benchmark across 49 repositories finds frontier coding agents rarely retrieve AI contribution rules on their own - and never refuse to contribute in AI-banned repositories, no matter the prompt. Disclosure and verification can be fixed; bans cannot.
6 min read
AGENTS.md Files Don't Move Coding Agent Correctness: A 288-Run Ablation
A controlled ablation across Claude Code and Codex, 17 real tasks, and 288 evaluated runs finds context-injection strategy does not measurably change correctness (bounded to under 10-15pp). The failures are implementation skill, not missing repository knowledge.
7 min read
LLMs Resolve Java Merge Conflicts Better Than Structured Tools - Because They Never Give Up
A calibrated study on real ConflictBench Java conflicts finds LLM agents match the developer's own resolution on 55-59% of true conflicts versus 36.7% for the best structured tool. The edge is coverage, not accuracy: the tools abstain on 20-90% of conflicts, the LLM on none.
6 min read
Kimi Linear: An Attention Architecture That Outperforms Full Attention
Moonshot AI's Kimi Linear paper introduces KDA, a hybrid linear attention that beats full attention at all scales - 75% less KV cache, 6x decoding at 1M context, and open-source checkpoints.
9 min read
GPT-5.6 Closes 30-Year Gap in Convex Optimization Theory
A researcher's 10-page domain-expert prompt helped GPT-5.6 produce a Lean-verified proof closing a complexity gap that stood since 1996. The paper is now on arXiv.
5 min read
GPT-5.6 Sol Ultra Produces Proof of the Cycle Double Cover Conjecture
OpenAI claims GPT-5.6 Sol Ultra has generated a proof for a 50-year-old graph theory conjecture in under an hour. The math community is now verifying whether it holds up.
5 min read
Anthropic Discovers J-Space: A Global Workspace Inside Language Models
Anthropic's new research reveals LLMs have an internal 'workspace' for silent reasoning - and it could change how we build safer AI.
8 min read
Leanstral 1.5: Mistral's Open Theorem-Proving Model Hits 100% on miniF2F
Mistral releases Leanstral 1.5, an Apache-2.0 licensed 119B parameter model (6B active) for Lean 4 theorem proving that saturates miniF2F and achieves SOTA on FATE benchmarks.
8 min read
VibeThinker-3B: A 3 Billion Parameter Model That Outscores Opus 4.5 on Reasoning
A new paper shows a 3B parameter model hitting 94.3 on AIME26 and 96.1% on LeetCode contests - matching or exceeding models 100x its size. The catch: it traded general knowledge for pure reasoning ability.
6 min read
Keep exploring
More on AI Research
- - Tools Directory - dive deeper across the Developers Digest knowledge base
- - All AI Research articles in the blog archive
- - Developers Digest on YouTube - video tutorials covering AI Research and more
Get Smarter About AI Dev
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.
One email per weekReal code, not theoryFree forever
Explore 862 topics
Browse All Topics