# The Frontier Wire

> The Frontier Wire covers artificial intelligence with news, research briefings, tool comparisons, industry surveys and editorials, written for people who need to understand what actually changed.

Full text of every article: https://thefrontierwire.com/llms-full.txt

## News

What happened in AI this week: model releases, funding, policy, and the people behind them. Index: https://thefrontierwire.com/news/

- [The reasoning-model era has a cost problem](https://thefrontierwire.com/articles/the-reasoning-model-era-has-a-cost-problem/): Models that think before they answer are better, slower and far more expensive per query. Here is what that shift means for anyone paying the bill.

## Research

Briefings on the papers, benchmarks and techniques that move the field, explained without the jargon. Index: https://thefrontierwire.com/research/

- [What test-time compute actually means, and why it changed the scaling debate](https://thefrontierwire.com/articles/test-time-compute-explained/): A plain-language explainer of test-time compute, the idea that spending more computation at inference can substitute for a bigger model.
- [Reading a model card like an auditor](https://thefrontierwire.com/articles/reading-a-model-card-like-an-auditor/): Model cards are the closest thing AI has to a nutrition label. Here is what to look for, what is usually missing, and how to tell a card from a brochure.

## Tools

Reviews and field notes on the models, frameworks and products people actually build with. Index: https://thefrontierwire.com/tools/

- [Field notes: running an AI gateway in front of every model call](https://thefrontierwire.com/articles/running-an-ai-gateway-in-front-of-every-model-call/): What an AI gateway does, what it costs you in latency, and the four features that turned out to matter after six months in production.

## Comparisons

Side-by-side evaluations of competing tools and models, with the criteria spelled out. Index: https://thefrontierwire.com/comparisons/

- [How to compare LLM evaluation frameworks without getting sold to](https://thefrontierwire.com/articles/how-to-compare-llm-evaluation-frameworks/): The seven criteria that separate evaluation tools in practice, and a scoring sheet you can fill in yourself before any vendor demo.
- [RAG vs. long context: choose by failure mode, not by benchmark](https://thefrontierwire.com/articles/rag-vs-long-context-choose-by-failure-mode/): Retrieval and long context windows solve the same problem in different ways. The right choice depends on how each one fails on your data.

## Editorials

Opinion from our editors and guest writers on where AI is going and where it should not. Index: https://thefrontierwire.com/editorials/

- [Stop calling it hallucination](https://thefrontierwire.com/articles/stop-calling-it-hallucination/): The word implies a malfunction. What the models do is closer to confident guessing, and the fix depends on naming it correctly.

## Surveys

Landscape surveys and state-of-the-field reports that map an entire category at once. Index: https://thefrontierwire.com/surveys/

- [The state of open-weight licensing, mapped](https://thefrontierwire.com/articles/the-state-of-open-weight-licensing-mapped/): Open-weight models ship under at least five kinds of licence, and only some of them are open in the usual sense. A field guide to what you can actually do.

## Site

- [Archive of every article](https://thefrontierwire.com/archive/)
- [Topics](https://thefrontierwire.com/topics/)
- [About and editorial standards](https://thefrontierwire.com/about/)
- [RSS feed](https://thefrontierwire.com/rss.xml)