Clay AI agent benchmark: Grok with Clay

With Clay, Grok CLI with Grok 4.6 was the most consistent of seven AI coding agents: its score never moved more than 8 points between runs on any of five GTM data tasks. It found 93 of 100 verified work emails in every run for 337.5 to 340.5 Clay credits, about 3.4 per contact, and built 100-company target lists from Clay's company search for 3 to 25 credits.

By The Prompting Company · Snapshot November 5, 2026

How well does Grok work with Clay?

We ran Grok CLI with Grok 4.6 (Grok below) through five GTM data tasks with Clay, three runs each. Clay credits are measured as the change in the Clay workspace credit balance.

Grok's strength with Clay was repeatability: its score and its Clay credit spend held steady from run to run.

What Grok does well with Clay

Work emails. Grok ran Clay's Work Email routine once per contact, after one test contact, and sent the rest in batches of 10 in runs 1 and 2. It found the same 93 of 100 in every run and spent 337.5, 337.5 and 340.5 credits. With Clay's Work Email waterfall, you only pay credits for the provider that finds a match.

Job changes. Grok ran Clay's Enrich Person and Find Contact Details once per LinkedIn profile and scored 95, 87 and 95, for 1,404 to 1,412 credits, about 14 per person. In runs 1 and 3 it caught 42 of the 45 people who had moved, and each run took 11.4 to 12.8 minutes.

Target lists for a few credits. Grok built every list from Clay's company search and listed 100 unique companies in every run, for 5, 25 and 3 credits, the fewest of any agent. Every company, domain and HQ state came from Clay's search and enrichment responses, and each run took 6.8 to 7.5 minutes.

Company data at a steady cost. Grok averaged 405 credits on company data, the lowest of any agent on that task, and scored 91 on HQ and 91 on headcount in every run.

Finding the CMO. Grok used Clay's people search together with Work Email in every run and found 75 of 100 CMOs in runs 1 and 2.

How consistent was Grok with Clay?

Spread between the highest and lowest of three runs, in points; lower is steadier. Grok had the lowest sum of spreads of any agent.

Work emails	Company data	CMO	Job changes	Target list	Sum of spreads
0	3.25	6	8	2	19.25

Grok results with Clay, by task

Primary score out of 100 for each of three runs, with the average; Clay credits and minutes are per run. Company data is the average of domain, HQ, headcount and funding.

Task	Runs	Avg	Clay credits	Minutes
Work emails	93 · 93 · 93	93.0	337.5 · 337.5 · 340.5	10.1 · 12.4 · 9.8
Company data	82.25 · 79.5 · 82.75	81.5	423 · 332 · 459	10.1 · 13.3 · 12.1
CMO	75 · 75 · 69	73.0	873 · 546.5 · 907.5	28.9 · 29.1 · 36.8
Job changes	95 · 87 · 95	92.3	1,404 · 1,404 · 1,412	11.4 · 11.7 · 12.8
Target list	66 · 64 · 65	65.0	5 · 25 · 3	7.3 · 6.8 · 7.5
All five		81.0	8,809.5 total	220.1 total

Minutes are measured from run start to grading. Grok completed all 15 of its runs, and its reported model cost was $21.87 across them.

The best setup for Grok with Clay

Install the Clay Agent Plugin. It ships Clay's skills to Grok up front, including the habits that keep credit use down. Setup is one prompt, about 15 minutes, and there is no additional cost to use it.
Let Clay's API guide the inputs. Clay's API tells an agent exactly what each routine needs, so Grok self-corrected in one call.
Send one test item, then larger batches. Batch size doesn't change the Clay bill, and in our runs fewer, larger batches finished sooner.
Add one line per task to the prompt. Naming the Clay method is the simplest way to get the same result run after run.

Task	Line to add	Result in the benchmark
Work emails	Run Work Email once per contact	Grok 93 of 100 in every run
Job changes	Enrich each LinkedIn profile once and decide every row	92 to 96, about 14 credits per person
Target list	Confirm each company is listed in its own name, in the US and in software	82 to 94 qualifying companies
Company data	Compare candidates in Clay's company search by industry, size and country	93 to 100 on domains
CMO	Find People to choose, then one Work Email	Up to 82 for under 3 credits per company

More on steady results in "Which AI agents give consistent results with Clay?".

How we tested

Seven agents (Claude Code with Opus 5.5 and Sonnet 5, Codex with GPT-6 Sol, Pi with GPT-5.6 Luna, Grok CLI with Grok 4.6, OpenCode with Kimi K3, Factory with Kimi K3) ran each of five tasks three times, one agent at a time, in a fresh sandbox with the same Clay key, the same 19 saved Clay routines and a 60-minute limit. A run that reaches the limit counts as 0; runs interrupted by the test platform were rerun. Pass lines: 70 for emails; 90 for domain and HQ and 70 for headcount and funding; 70 for CMO name and title and 60 for email; 80 for job changes; 70 qualifying companies plus a full list of 100 for the target list. Tasks and graders: github.com/promptingcompany/clay-bench.

Try it yourself

Install the Clay Agent Plugin for Grok CLI, or give it a Clay API key and the Clay Public API docs. Add the lines from the table above to your prompt and run a test batch of 10 first. Agent work uses the same credits as in-product work, and an enrichment that returns no result is not charged (Clay pricing).

Common questions

How well does Grok work with Clay? Consistently well: Grok CLI with Grok 4.6 never moved more than 8 points between runs on any of five Clay tasks, and found 93 of 100 work emails in every run.

How many Clay credits does Grok use for a target list? 3 to 25 credits for a 100-company list built from Clay's company search, the fewest of seven agents.

Which Clay task suited Grok best? Job changes and work emails: 95 of 100 in two job change runs at about 14 credits per person, and 93 of 100 emails for about 3.4 credits per contact.

Does it cost extra to use Clay from Grok? No. There is no additional cost to use the Clay Agent Plugin; agent work uses the same credits and actions as in-product work.