Gong Credits Explained (2026): How the New AI Usage Model Works

Gong credits are a usage-based AI metering system Gong rolled out in 2026. Each paid, core seat includes 2,000 pooled credits a year. Credits meter the capabilities that process data at scale — AI Trackers, MCP server tools, and API-based AI workflows. Core features like Gong Assistant, coaching, and forecasting stay included with your seat.

Quick answer: Gong credits are a company-wide pool — 2,000 per paid core seat per year — that meters only Gong's at-scale AI processing: AI Trackers, MCP server tools, and API-based AI workflows. Interactive, everyday use (Gong Assistant, coaching, forecasting, running a brief by hand) doesn't draw credits. AI Tracker configuration is the biggest consumption driver, so the main levers are filtering trackers, scoping queries tightly, and trimming brief sections. When the pool empties, credit-based requests error out and trackers pause until you add credits or the contract year resets.

Key takeaways

• ·Gong introduced Gong credits in June 2026, a usage-based layer on top of seat licensing that meters select AI processing rather than raising seat prices across the board.
• ·Each paid, core seat includes 2,000 pooled credits a year, shared across the company and reset at the start of each contract year.
• ·Credits meter AI Trackers, MCP server tools, and API-based AI workflows. Core features like calls and call analysis, coaching, forecasting, and Gong Assistant stay included with your seat.
• ·When the pool runs out, credit-based requests return an error and AI Trackers pause processing until you add credits or the contract year resets.
• ·AI Tracker configuration is Gong's own stated biggest consumption driver. Filtering trackers, using Deal-level queries instead of Account-level, and limiting brief sections are the main levers for controlling cost.

At a glance: what draws Gong credits

Gong facts checked against Gong's help documentation and Gong's own blog in July 2026.

What are Gong credits?

Gong credits are a new usage-based model Gong introduced to meter select AI capabilities. Gong describes the change as removing fixed limits on certain AI features and enabling workflows that process calls, emails, and customer interactions continuously and at scale. Rather than raising seat prices for every customer to support that expanded processing, Gong applies credits only to the capabilities that use it.

The rollout happened in phases through June 2026. Gong customers received an email in late May 2026 with company-specific credit allocations and expected usage. The official announcement page for the change is dated June 9, 2026.

Gong's co-founder and Chief Product Officer, Eilon Reshef, framed the model in a company blog post: credits are meant to track the actual work Gong's AI agents do on a customer's behalf, not a raw token count. A customer extracting structured data from a year of interactions consumes more credits than one working with a single month, because more data gets processed. Gong's stated goal is to price that expanded automation comparably to, and often cheaper than, running the same processing through a third-party model API. That's Gong's own framing, worth validating against your actual usage rather than taking as a given.

How many credits do you get, and how do they work?

Each paid, core seat includes 2,000 credits annually. Credits are pooled and shared across the whole company rather than assigned to individual users, and the pool resets at the start of each contract year.

For the initial rollout, Gong also gave companies a one-time introductory credit allocation, meant to help teams get familiar with how their usage draws down the pool before the standard allocation took over. That introductory bump applies only to this first transition, not every contract year going forward.

If a company needs more than its included allocation, it can purchase additional credits. Purchased credits are available until the end of the current contract term and do not roll over into a new one. Gong does not automatically add credits or generate charges. Buying more is a deliberate choice a company makes, not something that happens by default when usage climbs.

Admins can monitor credit balance and consumption trends from Admin Center, under the Gong credits page in Settings. From there, admins can download credit usage reports for a chosen date range. Gong also sends in-app notifications and admin emails once usage crosses defined thresholds, so a company isn't finding out it's low on credits only after something stops working.

What consumes Gong credits?

Three categories draw from the credit pool: AI Trackers, Gong's MCP server, and API-based AI workflows. Gong has also said it's extending the same model to pre-built agents like AI Call Reviewer and AI Data Extractor.

AI Trackers are, in Gong's own words, the biggest lever on credit consumption. Part of why: Gong recently introduced question-based AI Trackers, which let teams spin up a tracker by describing what they want in natural language instead of manually training it with examples. That made trackers dramatically easier to create. Reshef's blog post notes that in the few months after that feature shipped, customers created more AI Trackers than in the previous four years combined. More trackers, especially broad ones scanning every company conversation, means more ongoing credit draw.

MCP server tools consume credits every time they run. The ask_account and ask_deal tools each analyze the calls and emails in scope for that request, and they redo that analysis on every call, so asking the same question about the same account twice analyzes the same conversations twice. The generate_brief tool is similar but multiplies by section: each open-ended section in a brief triggers its own analysis pass, so a ten-section brief analyzes the underlying conversations ten times over.

Automated and scheduled use compounds this. A brief generated through the Gong interface by a person doesn't consume credits, but the same brief generated on a recurring schedule through the API or MCP does, every single time it runs, whether or not anyone reads the output.

What's included with your seat, no credits required?

Several core AI capabilities stay included with your existing seat and don't touch the credit pool: calls and call analysis, conversation insights, deal intelligence, forecasting, coaching, and revenue and deal predictions. Gong Agents used interactively by individual users are also included.

Gong Assistant, the interactive chat interface replacing AI Ask Anything, is explicitly covered by your seat license. Gong's own blog post is direct about this: everyday user-facing tasks like writing and rephrasing emails, prepping for meetings, and using Gong Assistant don't draw credits, so that only advanced, at-scale automation does.

The same logic applies to briefs. Running one manually inside the Gong interface is free; automating it through the API or a schedule is not. And AI Theme Spotter sits outside the credit system entirely. It runs on its own separate monthly quota (capped between 10 and 30 analyses per company per month, depending on seat count), a mechanic that predates and doesn't overlap with Gong credits.

What happens when you run out of credits?

Behavior differs slightly by capability. For API and MCP requests, Gong's documentation is direct: if there aren't enough credits to complete a request, the API returns an error. For AI Trackers, processing pauses instead. Calls stop being analyzed by trackers that have exhausted the pool.

Unpublishing a tracker stops its credit consumption if you want to cut usage proactively. On the recovery side, adding credits automatically resumes processing for paused API and MCP calls. AI Trackers, however, need to be manually resumed, and once resumed they'll process the calls that piled up while paused.

To keep processing past your included allocation, you purchase additional credits through your Gong account team. Nothing about your existing seat-based agreement changes when this happens; credits are an addition to that agreement, not a replacement for it.

How do you reduce Gong credit consumption?

A handful of practical levers show up repeatedly in Gong's own optimization guidance:

Narrow your date ranges. A broad or unspecified range on an MCP request can pull in hundreds of emails and dozens of calls for a single account. Defining a specific window, like the last 30 days or the current quarter, cuts consumption without hurting answer quality in most cases.

Query at the right entity level. Account-level queries analyze conversations across every deal and contact tied to that account. Deal-level queries are scoped to a single opportunity. If the question is really about one deal, ask it at the deal level.

Combine related questions. Each MCP request re-analyzes the selected conversations from scratch. Four separate questions about the same account can mean the same calls get analyzed four separate times. Bundling related questions into one request cuts that down, though Gong cautions against combining unrelated questions just to save credits, since that tends to produce a less focused answer.

Base brief sections on existing trackers where you can. A brief section built on data an AI Tracker already surfaced doesn't require re-analyzing the underlying calls. A section that processes calls or emails directly does, every time the brief runs.

Trim brief sections to what's actually needed. Since each open-ended section is a separate analysis pass, a shorter, more targeted brief consumes meaningfully less than a comprehensive one built for every possible use case.

Skip web-search sections unless you need them. Sections that pull in external information alongside Gong data tend to consume more credits than sections that only use Gong's own data.

Filter and prune AI Trackers. Scope trackers by team, user, account type, or deal stage instead of scanning every company conversation, and unpublish trackers that have stopped providing value.

Generate on demand, not just in case. Pre-generating briefs or Ask responses for accounts nobody may ever review burns credits for output that never gets used.

Does Discera use your Gong credits, or need an extra seat?

No, on both counts, and the reason is the same for each. Gong scopes credits to its own AI agents doing background work: AI Trackers, MCP tools, and API-based AI workflows. Plain data retrieval through Gong's standard API isn't one of those. Discera connects to Gong read-only through OAuth and uses Gong's basic call and transcript endpoints — the same non-credit-metered part of the API that any integration uses to pull raw data — to retrieve your calls. It stops there. Discera never calls Gong's AI Trackers, MCP server, or Ask Anything.

Everything after that retrieval — running your custom questions against each call, extracting findings, aggregating them into a report — happens entirely on Discera's own infrastructure. Gong never does any of that processing, so there's nothing for Gong to meter. That's also why Discera doesn't require an additional Gong seat for the connection itself: one Gong admin authorizes the OAuth connection once, and Discera pulls data from there without occupying a license the way an interactive Gong user would.

Practically, that means a company already running close to its Gong credit ceiling doesn't add pressure to that ceiling by connecting Discera, and doesn't need to buy another Gong seat to do it either. Worth confirming against your own Gong credit usage report after connecting, since the specifics of what counts as which kind of API call ultimately depend on Gong's live product behavior, not just its public docs.

For a full breakdown of how Discera and Gong's native AI compare on cost, recurrence, and non-seat access, see Discera vs Gong native AI, or the broader comparison of Discera against Gong's full AI suite. On the seat question specifically, how to give teams Gong access without a seat goes deeper. Discera's own pricing has no credit system and no per-seat charge to manage: it's flat, based on analyzed call volume, starting at $29/mo on the Starter plan.

Try Discera free

100 calls, 30 days, no credit card. First report in minutes. Start your trial at app.discera.ai/sign-up

Design partner program is open. 10 slots, full access free for 3 months, founder-direct, real influence on the roadmap. Apply for a design partner slot