#ai-procurement

The NSA, CISA and FBI name six AI companies in advisory AA26-251A. See what changes for model procurement, provenance checks and API abuse monitoring.

Turn an impressive AI demo into a buying decision. Ask for buyer-chosen tasks, repeat trials, visible interventions and a usable result before committing.

An anonymous model's benchmark score and its identity are different products. What stealth-listing terms disclose, and what no eval can tell you.

We collected 42 vendor-published benchmark rows from 9 vendors and checked what each page discloses: version, subset, harness, effort. Full table in the post.

Price-change notice is a contract term, not a courtesy, and the periods range from a fortnight to none at all. What to check before committing model spend.

Three Chinese labs shipped major releases and their licence stories diverge. A buyer's scoreboard on weights, pricing schedules, and benchmark claims.

Time-of-day LLM pricing is arriving, and a cheaper window is not the same as a cheaper bill. How the announced schedules and peak hours actually work.

A catalog row can show a real, live price that is a time-boxed promotion, a surface-specific rate, or a listing date that is not a launch date at all.

Four levers decide what a vendor benchmark table means: which benchmark version, which task subset, what shape the metric takes, and who got a column.

A signing-stage checklist for AI contracts: training-data clauses, retention, model deprecation, usage limits, sub-processors, exit and export rights.

Mistral is the only frontier-class vendor whose EU answer is structural: Paris HQ, announced EU inference capacity, open weights. When the premium pays.

Oracle customers can soon apply Universal Credits to OpenAI models and Codex via OCI. Why this procurement-rails shift matters for regulated enterprises.

OpenAI's confidential S-1 lands as Anthropic heads public too. The first time buyers have two public AI vendors, and why you should design for portability now.

Stage 4 of the agentic AI pipeline — vendor scorecards, RFP, evaluation rubric, and the templates that turn vendor pitches into defensible decisions.

Why $/token is the wrong unit and $/successful-task is the right one. Formulas, worked examples across 6 task families, and a downloadable scoring template.