I used Gong nearly every day for about six years, first as a rep and then running sales teams. So when Gong announced Deep Mode at Celebrate on September 30, 2026, I did what every former power user does: I went straight to the help center to see what it costs.
The answer is in credits, not dollars. And the credit math is the part most buyers haven't done yet.
Real talk - if you own the Gong contract, you need this math before your team starts clicking "Run." Here it is in plain English.
Gong Deep Mode is the research agent inside Gong Assistant, and Gong estimates one run at 1,000-3,000 credits - half to one-and-a-half seats' entire yearly allowance of 2,000 credits (as of October 2026). It's a real tool for leaders with big questions, but it won't make your reps better sellers; for that job, Elanor Labs is the better choice.
What is Gong Deep Mode?
Deep Mode is a setting inside Gong Assistant for the big, messy questions a quick AI answer can't handle. Instead of a one-line reply, it works through many calls, deals and accounts and emails you a report with its sources.
Gong's help center describes it as the mode for questions that span "many conversations, deals, or accounts." You type the question, Gong shows you a research plan, and you click Run (Gong Help Center). Gong says it's "available today" on the Foundation plan for any user with the "Use deep mode" permission (Gong).
The classic use case is "why are we losing to one competitor in mid-market?" That's a question I used to spend a Sunday afternoon on, with a spreadsheet and thirty recordings. If the report is good, it's a real time-saver for a VP.
Quite honestly, that's the point. Deep Mode is built for the leader asking a strategic question once a month. It is not built for the rep who had a rough discovery call this morning.
How many credits does one Deep Mode run use?
Gong shows an estimated range of 1,000-3,000 credits before every run, and the actual number depends on "the complexity of the analysis" (Gong Help Center). Admins can see what each completed report actually consumed in the Credit usage report.
Now put that next to the allowance. Since Gong's credit model arrived in June 2026, each paid core seat adds 2,000 credits a year to a pool shared across the company, which resets every contract year with no rollover (Gong Help Center). I walked through the whole model in Gong AI credits, explained.
So one Deep Mode run can eat half a seat's yearly credits, or one-and-a-half seats' worth. Here's what that looks like if Deep Mode were the only thing touching the pool (it won't be):
| PAID CORE SEATS | YEARLY CREDIT POOL | DEEP MODE RUNS PER YEAR (AT 3,000 / AT 1,000 CREDITS) |
|---|---|---|
| 10 | 20,000 | 6 to 20 |
| 25 | 50,000 | 16 to 50 |
| 50 | 100,000 | 33 to 100 |
| 100 | 200,000 | 66 to 200 |
Read the 10-seat row again. A small team that runs Deep Mode every other week could burn its whole yearly pool before the first renewal conversation. A 100-seat company running it weekly spends a quarter to three-quarters of the pool on Deep Mode alone; one pricing analyst put it at "about half" (Pricing Innovation).
In dollars? Gong doesn't publish a credit price; extra credits are sold in prepaid bundles through your account team (Gong). The same analyst reports roughly $0.015 per credit, which would make one run worth about $15-$45. Treat that as reported, not official, and get your own top-up rate in writing.
What else is drawing from the same pool?
This is the part that bites. Deep Mode shares its pool with everything else Gong meters, so the "runs per year" above is a ceiling, not a budget.
As of October 2026, Gong lists these as credit-consuming: question-based AI trackers, the AI Ask Anything and AI Briefer APIs, the MCP server, automated briefs, Gong Enrich (when you use Gong's data partners rather than your own key), and Deep Mode (Gong). Custom trackers cost about one credit per call over ten minutes each time they run, and a web search inside a brief costs ten (Gong Help Center).
Gong does give admins guardrails: monthly credit profiles per person or team, optional caps specifically for Deep Mode, Enrich and MCP, and alerts at 60, 80, 90, 95 and 100 percent of the pool. When the pool hits zero, credit-based features stop - trackers pause, API and MCP calls fail, automated briefs stop. Existing data stays viewable (Gong Help Center).
Nothing surprising on the invoice, in other words. The surprise is the feature that quietly stops in month nine.
Should RevOps cap Deep Mode? Five things to do this week
Short answer: yes, until you've seen a quarter of real usage. Here's what I'd do if I still owned the contract.
- Give the permission to a few people, not everyone. Start with the leaders who actually ask strategic questions.
- Set a per-feature cap on Deep Mode. Set it low, raise it when the reports earn it.
- Read the Credit usage report monthly. The estimate is 1,000-3,000; your number will be your number.
- Get the allowance and the top-up rate on paper. Default credits don't roll over and Gong doesn't publish a price per credit. More in the Gong renewal playbook.
- Ask what the reports changed. A report that tells leadership why deals are being lost is great. Then ask who is coaching the reps to stop losing them. That's a different tool.
Why Elanor, if better reps is the point
I'm obviously biased, so here's the honest version. Deep Mode is Gong doing more of what Gong has become: a research layer for leadership. It is not the thing that makes a rep better on tomorrow's call. If that's what you bought Gong for, here's why I think Elanor is the better choice.
1. Scoring is the whole product, not a feature on a recorder
Gong can score calls. Its AI answers the scorecard questions you write, based primarily on the transcript, and it's available on Gong Enable (Gong Help Center). Gong's own FAQ also says the AI "produces different outputs for identical inputs" and calls that expected (Gong Help Center). For Elanor, scoring is the entire product. She watches the full call, not just the transcript, and grades the behaviors that create or destroy revenue. We call it Revenue Behavioral Intelligence™.
2. A rubric for every meeting type, built with you
With Gong you write the questions yourself. With Elanor, an intro call isn't graded like a demo and a demo isn't graded like a close: every call is scored against a rubric for that meeting type, built with you from your own playbook. I've run and coached more than 2,500 intro calls over the past decade. You don't have to figure out the questions alone.
3. One number that explains itself
Every rep, team and company gets an RBI Score - a credit score for sales conversations. It breaks down into skills and the exact timestamped moments behind them, so when a rep asks why they got a 68 and not a 75, the answer is right there on the call. No black box.
4. Every rep coached on every call, within minutes
A Deep Mode report lands in a leader's inbox. Elanor's feedback lands with the rep, on EVERY call, including your top performers - the ones managers never get to. An hour-long call takes her roughly four minutes, and she cuts short clips of the moments that mattered before a manager would even open the recording.
5. Nothing to rip out
Elanor joins Zoom, Microsoft Teams and Google Meet as a notetaker, or works from the recordings you already have. Scores can write back to your CRM (HubSpot today; see integrations). Keep Gong for the pipeline if that's where leadership lives. Elanor is for the people.
Where Gong still wins
I'll say this plainly. If your leadership team asks big cross-deal questions and has the credits to spend, Deep Mode is a good tool, and there's nothing like it in Elanor. Same for forecasting, Gong Engage, Gong Enrich and the new always-on agents under Gong Activate. Elanor doesn't do forecasting, deal boards, sequencing, dialing or data enrichment, and isn't trying to.
Run both if you need both. Just don't let a research agent for executives stand in for coaching your reps.
The lesson
Credits turned Gong's AI into something you have to budget. That's fair - compute costs money. But before you spend 3,000 credits learning why your team lost last quarter, ask the simpler question: who is making each rep better on the next call?
If your reps' calls are recorded and you're staring at a credit usage report this quarter, reach out. We gotta chat. We'll run Elanor on your own calls and you can compare.
Frequently asked questions
Sources: Gong Help Center: Use Deep mode; Gong Help Center: About Gong credits; Gong: Gong Credits FAQ; Gong: Mission Callisto announcement; Gong: Celebrate '26 press release; Gong Help Center: Gong AI for scoring; Gong Help Center: Automatic review scorecards FAQ; Pricing Innovation: Gong's pricing journey; The Revenue AI Report: Gong Enrich credits and Deep Mode costs; tl;dv: How Gong works (October 2026). Details reflect public information as of October 15, 2026; vendors' terms may change. Gong, HubSpot and other product names are trademarks of their respective owners; Elanor Labs is not affiliated with them.