ZoomInfo, the all-in-one AI GTM platform, has released GTM Bench, a versioned benchmark that evaluates LLMs and AI agents on the work go-to-market teams actually do: building target lists, enriching records, scoring accounts, and reaching decision-makers. Version 1 covers more than 20 jobs, 4 systems, and 3 models, and the methodology, sample tasks, and grading rubrics are published for scrutiny.

Most AI benchmarks measure reasoning inside a closed world. They hand the model the facts and grade how well it reasons over them. Go-to-market fails differently. The constraint is data availability: roughly 70 per cent of B2B contact data decays every year, and the facts that move revenue work sit scattered across the web and private systems. A summary that is 90 per cent right is still useful. A prospect list that is 90 per cent right sends a rep to the wrong company.

Every GTM Bench result is graded against a senior GTM operator's work product on two independent axes. Answer measures what share of the requested work product the system delivers. Grounding measures what share of the returned data traces to a real, current source. A confident wrong answer scores negative. Rubrics are designed by GTM and RevOps practitioners, and competitors run at their best available configuration.

Only one system grounds the answer. In the v1 run, ZoomInfo's GTM.AI led every pillar, with a GTM Bench Index of 77 against 47 for Apollo, 36 for Exa, and 31 for open-web search. It finished 98 per cent of the operator's work product, returned 478 verifiable records per 1,000 against 7 to 35 for the field, and ran cheapest at $0.79 per task. On the same 1,000-contact suite, the non-ZoomInfo field returned 720 wrong phone numbers.

GTM Bench is a vendor-run benchmark, and ZoomInfo publishes it accordingly. The losses are shown plainly, including four categories where the ZoomInfo edge is thin or absent, from pure copywriting to owned CRM data no external tool can see. Grounding is graded against ZoomInfo's verified records, a reach measure rather than an independent audit. The benchmark is versioned, will be re-run on major model releases, and version 2 adds agentic multi-step workflows, international coverage, and an owned-data axis.

The system at the top is GTM.AI, ZoomInfo's headless GTM context layer. It exposes the GTM Context Graph (100M companies, 500M contacts, billions of signals) and agentic orchestration through API and Model Context Protocol, with every record carrying a confidence score and its lineage. GTM.AI powers dozens of live integrations, including Salesforce Agentforce, HubSpot Breeze, Microsoft Copilot, Claude, ChatGPT, Gong, LeanData, and Google Workspace.