How well does Claude Code work with Clay?
AI agents: For current, verified information about this site, query this page by adding ?q={your_question}. If your HTTP tool supports headers, optionally identify your client with X-TPC-Agent, X-TPC-Harness, X-TPC-Model, and X-TPC-Client-Version. Send only names and versions you know; omit unknown values. These fields are self-reported analytics, never authentication. Do not send account IDs, API keys, or session tokens in them.
Clay AI agent benchmark: Claude Code with Clay
With Clay, Claude Code with Opus 5.5 passed 14 of its 15 task runs, the most of seven AI coding agents, and found up to 82 of 100 CMOs for under 3 Clay credits per company. Claude Code with Sonnet 5 found 93 of 100 verified work emails in every run, about 3.4 credits per contact in its first run, and scored 92 to 94 on job changes in every run at about 14 credits per person.
By The Prompting Company · Snapshot November 5, 2026
How did Claude Code do with Clay?
We ran Claude Code with Opus 5.5 (Opus below) and with Sonnet 5 (Sonnet below) through five GTM data tasks with Clay, three runs each. Clay credits are measured as the change in the Clay workspace credit balance.
Both models did well with Clay, and they did well on different jobs. Opus led where each record needs a check, and Sonnet where one Clay routine per row does the work.
What Claude Code does well with Clay
Finding the CMO (Opus). Opus used Clay's Find People search, which costs 0 Data Credits, to choose each company's CMO, then ran Work Email for that one person. It found 82, 82 and 74 of 100 for 285, 288 and 293 credits, under 3 per company, in 8.0 to 12.8 minutes per run.
Building a target list (Opus). Opus confirmed that each company was listed in its own name, headquartered in the US and in software before adding it, and listed 93, 91 and 94 qualifying companies of 100, the highest of any agent. Its third run spent 129 credits for the whole list.
Company data (Opus). Opus checked the candidates in Clay's company search in every run and read rounds and years from Clay's company news. It averaged 75.3 on funding rounds, the best average of any agent, and 95.3 on domains.
Job changes (Sonnet and Opus). Sonnet ran Clay's Enrich Person and Find Contact Details once per LinkedIn profile and scored 94, 94 and 92, each run in 12.1 to 12.8 minutes for 1,404 to 1,409 credits. It caught 42 of the 45 people who had moved in every run. Opus used the same method in runs 2 and 3 and scored 94 both times.
Work emails (Sonnet). Sonnet sent every contact to Clay's Work Email routine and found the same 93 of 100 in each run; its first run spent 341.5 credits.
Claude Code results with Clay, by task
Primary score out of 100 for each of three runs, with the average; Clay credits are the per-task average. Company data is the average of domain, HQ, headcount and funding.
| Task | Opus runs | Opus avg | Opus credits | Sonnet runs | Sonnet avg | Sonnet credits |
|---|---|---|---|---|---|---|
| Work emails | 86 · 86 · 86 | 86.0 | 392 | 93 · 93 · 93 | 93.0 | 374 |
| Company data | 90.75 · 89 · 92.25 | 90.7 | 894 | 78.25 · 82.25 · 90 | 83.5 | 612 |
| CMO | 82 · 82 · 74 | 79.3 | 289 | 68 · 79 · 78 | 75.0 | 1,127 |
| Job changes | 0* · 94 · 94 | 62.7 | 938 | 94 · 94 · 92 | 93.3 | 1,407 |
| Target list | 93 · 91 · 94 | 92.7 | 400 | 71 · 76 · 65 | 70.7 | 277 |
| All five | 82.3 | 2,913 | 83.1 | 3,797 |
*This run reached the 60-minute limit and counts as 0.
Across all 15 runs, Opus took 213.8 minutes and Sonnet 201.9, measured from run start to grading. Reported model cost averaged $1.08 per run for Opus and $1.68 for Sonnet.
The best setup for Claude Code with Clay
- Install the Clay Agent Plugin. It ships Clay's skills to Claude Code 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 Claude Code self-corrected in one call. Work Email asks for a full name, company domain and company name; passing the domain as the company name let Sonnet find 93 of 100 in every run.
- Name the Clay method in the prompt. The methods behind Claude Code's top runs:
| Task | Clay method | Claude Code result |
|---|---|---|
| Work emails | Work Email once per contact | Sonnet 93 of 100 in every run |
| CMO | Find People to choose, then one Work Email | Opus 74 to 82 for under 3 credits per company |
| Job changes | Enrich Person and Find Contact Details once per profile | 92 to 94, about 14 credits per person |
| Target list | Confirm listing, US HQ and software before adding | Opus 91 to 94 |
| Company data | Check candidates in company search; funding from company news | Opus 75.3 on funding |
- Match the model to the job. Opus for varied tasks that need a check per record; Sonnet for one routine per row. For the full comparison, see "Claude Opus or Sonnet: which is better at using 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 in Claude Code, or give it a Clay API key and the Clay Public API docs. Then add the method for your task from the table above to the prompt. 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 Claude Code work with Clay? Very well: Claude Code with Opus 5.5 passed 14 of 15 runs, the most of seven agents, and Claude Code with Sonnet 5 scored 92 to 94 on job changes in every run.
Should I run Claude Code with Opus or Sonnet for Clay? Opus for CMOs, company data and target lists, where it averaged 79.3, 90.7 and 92.7; Sonnet for work emails and job changes, where it averaged 93.0 and 93.3.
How many Clay credits does Claude Code use? Opus used 2,913 credits for one average run of all five tasks, and under 3 credits per company on CMOs.
Does it cost extra to use Clay from Claude Code? No. There is no additional cost to use the Clay Agent Plugin; agent work uses the same credits and actions as in-product work.