Billing
ChatBotKit’s billing system is grounded in a simple idea: you only pay for the computational resources your bots use. To make that transparent, the platform converts every token generated or consumed by a model into a single bookkeeping unit called a “credit token.” The conversion hinges on a model-specific multiplier - its token ratio - so usage is predictable no matter which underlying model you choose. Below is an in-depth look at how cost is computed, how the standard subscription tiers work, and how organisations can evolve from an off-the-shelf plan to a bespoke “unlimited” agreement.
How Usage Cost Is Calculated
Every model on ChatBotKit has a published token ratio that reflects its relative cost. When your bot processes text or images, the raw model tokens are multiplied by that ratio to determine the number of credit tokens recorded for billing purposes.
This ledger-style approach means:
- Predictable spend across models: You can switch models mid-project and still understand the impact on budget immediately.
- Straightforward cost optimisation: Choose a lower-ratio model for high-volume tasks and reserve premium ratios for specialised reasoning or vision requests.
- Built-in transparency: Every plan advertises a monthly credit-token allowance, so you can map workloads to a subscription tier with confidence.
Sample Token Ratios
Model | Intended Strength | Token Ratio |
---|---|---|
gpt-4o-mini | Cost-efficient reasoning & vision | 0.0333 |
o3 | Balanced, powerful across domains | 0.4444 |
gpt-4 | Flagship broad-knowledge model | 3.3333 |
claude-4-opus | Top-tier coding & long tasks | 4.1667 |
gpt-4.5 | Highly creative, open-ended dialogue | 8.3333 |
Lower ratios mean cheaper per-token billing; higher ratios reflect premium compute or specialised capabilities. For more information see our Models documentation.
Standard Subscription Plans
ChatBotKit publishes several ready-made plans aimed at rapid self-service onboarding. Each tier bundles a monthly pool of credit tokens plus fixed caps on other resources (datasets, messages, crawled pages, etc.). These caps are designed to match the typical scaling path of teams moving from prototype to production.
Plan | Price / Month | Monthly Credit Tokens | Conversations | Messages | Notable Extras |
---|---|---|---|---|---|
Free | $0 | “Generous” starter allocation | Limited | Limited | Ideal sandbox for experimentation |
Basic | Lower than Pro | Modest pool | Limited | Limited | Entry-level exploration without long-term commitment |
Pro | $65 | 2 million | 10,000 | 50,000 | Unlimited bots, priority support, advanced crawling (1,000 pages), BYO models |
Team | $365 | 10 million | 50,000 | 250,000 | Premium support, 50,000-page crawls, prioritised feature development |
Enterprise | Custom | ∞ (Unlimited) | ∞ | ∞ | Dedicated account manager, custom SLAs, bespoke security & retention policies |
Every paid tier includes:
- Access to all platform AI models on preferential lanes.
- The ability to bring your own model and keys (BYOK) for cost control or compliance.
- End-user privacy enablement and brandable widgets to match your corporate identity.
Credit-Token Math in Action
Suppose your Pro-tier bot generates 600,000 tokens with gpt-4o-mini in a month. The recorded usage is:
600,000 tokens × 0.0333 ratio = 19,980 credit tokens
That is well below the 2 million-credit-token allocation - leaving ample headroom for additional queries, file uploads, or dataset expansions.
Why Start with a Standard Plan?
- Instant activation: Self-signup, select Pro or Team, and start building in minutes - no procurement cycles required.
- Baseline metrics: The platform’s analytics reveal your true traffic shape, message length, and model mix within the first few weeks.
- Predictive budgeting: Because the token ledger updates in real time, you can forecast burn and adjust temperature settings, context windows, or model choices before hitting tier limits.
These characteristics are intentional: they lower the barrier to entry while simultaneously giving ChatBotKit the data required to craft a future bespoke agreement.
Graduating to a Custom “Unlimited” Agreement
ChatBotKit’s philosophy is “try, measure, customise.” The move from a catalogue plan to an enterprise contract follows a three-step lifecycle:
- Trial Phase – The customer launches on a standard tier (often Pro or Team) and runs realistic workloads.
- Data-Driven Proposal – Usage telemetry - token mix, request frequency, peak concurrency - feeds into an internal pricing model. ChatBotKit then drafts terms optimised for the client’s exact pattern.
- Unlimited Plan Activation – The customer signs a bespoke agreement with uncapped resource ceilings, negotiated SLAs, and tailored data-governance clauses. In effect, the client now sits atop an “unlimited” tier backed by enterprise support and predictable monthly or annual billing.
Because credit-token accounting remains in place behind the scenes, ChatBotKit can still audit resource consumption, but the customer no longer sees soft caps - they enjoy elastic headroom aligned to contractual commitments.
Advantages of the Bespoke Model
- No surprise throttling during traffic spikes or campaign launches.
- Volume discounts baked into the flat fee, reflecting economies of scale.
- Regulatory alignment through custom data-retention windows, regional isolation, and security policies.
- Priority roadmap influence - feature requests get fast-tracked as part of the partnership.
Optimising Your Spend: Practical Tips
Choose the Right Model per Task
- Use gpt-4o-mini for high-volume conversational flows at 0.0333 ratio.
- Reserve claude-4-opus or gpt-4.5 for occasional complex reasoning when the higher ratio is justified.
Keep Context Windows Lean
- Truncate irrelevant history and attachments.
- Break long documents into smaller segments to avoid paying token ratios on redundant text.
BYOK for Specialised Workloads
If your in-house OpenAI key is discounted, plug it in via the BYOK interface. ChatBotKit will still meter credit tokens, but your OpenAI bill stays under your direct contract terms.
Monitor the Dashboard
Real-time analytics flag unusually large prompts, runaway loops, or elevated error rates that might translate to wasted tokens.
Frequently Asked Questions
-
“What if I exceed my credit-token pool?”
ChatBotKit notifies you and offers instant top-ups or a plan upgrade. Pro and Team users can also set hard stop-limits to prevent overruns.
-
“Can I host ChatBotKit on-prem?”
Yes. Enterprise agreements can stipulate a fully managed private deployment or an on-premise installation - pricing is adjusted accordingly.
-
“Do you offer annual discounts?”
All plans, including custom ones, can be billed annually for a reduced effective monthly rate.
Key Takeaways
- Credit tokens are the universal currency: model tokens × token ratio.
- Standard plans (Free, Basic, Pro, Team) provide quick, predictable onboarding with clear caps and published prices.
- Enterprise customers graduate to an unlimited plan after a metrics-driven trial, eliminating resource ceilings while locking in customised terms.
- Cost control remains actionable through model selection, prompt engineering, and BYOK strategies, even on unlimited contracts.
By blending transparent token accounting with flexible subscription pathways, ChatBotKit ensures that solo developers and Fortune 500 enterprises alike can scale conversational AI without budget surprises or capacity barriers.