Category / T-CAT / RESEARCH
Research
Research / T-2026-5713
Migrating to GPT-5.6: 2.2x faster, 27% cheaper, and a masterclass in hidden model debt
Ploy's migration from Claude Opus to GPT-5.6 Sol reveals how much of 'the model' is actually provider-specific infrastructure — and why every frontier model switch is a stack…
SourceMigrating a production AI agent to GPT-5.6: 2.2x faster, 27% cheaper
Research / T-2026-9928
The Four Horsemen Behind Postgres Outages: A Database Autopsy for the AI Age
Postgres outages plague AI startups. A Rust reimplementation, pgrust, targets four root causes: VACUUM, connection limits, bad query plans, and JSON.
SourceThe four horsemen behind Postgres outages
Research / T-2026-6314
A Public Ledger of Cloud Outages and the SLA Credits They Trigger
A new site tracks cloud outages and the SLA credits they trigger, turning vendor opacity into a structured data set for AI builders.
SourceA public ledger of cloud outages and the SLA credits they trigger
Research / T-2026-9444
Mesh LLM turns spare GPUs into a distributed AI cluster
Mesh LLM pools GPUs across machines into one OpenAI-compatible API. The project targets teams that want control over models, data, and hardware costs.
SourceMesh LLM: distributed AI computing on iroh
Research / T-2026-8805
Apple sues OpenAI, alleging a coordinated campaign of trade secret theft
Apple filed a lawsuit accusing OpenAI of orchestrating the theft of trade secrets via former employees, including a VP of product design and a senior engineer.
SourceApple sues OpenAI, accuses ex-employees of stealing trade secrets
Research / T-2026-7586
False Sharing Is an AI Infrastructure Problem: Rust's 128-Byte Rule
Why Rust's 128-byte alignment rule for multi-core structs matters for AI inference servers, training orchestrators, and the next generation of high-throughput infrastructure.
SourceCache-Conscious Data Layout in Rust: Field Zoning, False Sharing, 128-Byte Rule
Research / T-2026-9693
GPT-5.6: The efficiency play that changes the frontier math
OpenAI launches GPT-5.6 Sol, Terra, and Luna. The story is not the benchmark scores. It is the cost curve.
SourceGPT-5.6
Research / T-2026-1114
Frugon shows developers exactly which LLM calls are burning cash
Rodiun's open-source Frugon analyzes your LLM call logs locally to recommend cheaper models for easy tasks, promising 37% savings on a $549 monthly bill.
SourceShow HN: Frugon – Find which LLM calls a cheaper model could handle (local, MIT)
Research / T-2026-1403
Microsoft's Flint Rethinks Visualization as a Compiler Problem for AI Agents
Microsoft Research releases Flint, a visualization intermediate language designed to make AI agents reliably generate polished charts from simple specs.
SourceShow HN: Microsoft releases Flint, a visualization language for AI agents
Research / T-2026-4673
Fortress patches Chromium's C++ to make bot detectors see a human
Fortress patches Chromium's C++ to make bot detectors see a human. An open-source engine fork corrects the browser fingerprint at the native-code layer, clearing Cloudflare…
SourceShow HN: Fortress – a stealth Chromium so your agents stop getting blocked
Research / T-2026-6858
GLM 5.2 and the coming AI margin collapse
An open-weight model from Z.ai matches frontier quality at 20% of the price. The AI industry's business model relies on inference margins that may not survive.
SourceGLM 5.2 and the coming AI margin collapse
Research / T-2026-4048
The AI industry is learning the wrong lessons from Vasily Grossman
A Baffler essay on Vasily Grossman's Soviet epic carries an uncomfortable mirror for AI builders who mistake scale for meaning.
SourceThe Music of Destruction