Last Updated on June 10, 2025
22.1 What Are Tokens
22.1 What Are Tokens
A “token” is a unit of measurement used to quantify the amount of text or visual data processed by Chariot’s AI systems during a prompt, upload, or analysis. Tokens apply to both inputs (what you send) and outputs (what the AI returns), and they form the basis of usage tracking across features.
How Tokens Work
In text prompts, a token typically equals a few characters, part of a word, or punctuation mark
Example: “The car is red.” = approximately 5–6 tokens
In vision analysis, a token may correspond to scanned image segments, OCR regions, or semantic object groupings
Example: A clear photo of a car interior may consume 100–300+ tokens depending on complexity
In document analysis (PDFs), each parsed sentence or clause contributes to total token usage
Scanning 5 pages of legal contracts may use 2,000–5,000 tokens, depending on layout and formatting
Why Tokens Matter
Subscription plans are structured around daily, weekly, or monthly token caps
Chat prompts, image uploads, and PDF scans all deduct tokens
Token usage determines access levels, processing time, and billing enforcement
Token Limits
Each feature may define its own limits:
Single prompt: capped at ~4,000–8,000 tokens
Total session: capped at ~8,000–16,000 tokens
Subscription plans: capped at a certain number of tokens per day/month
Going beyond your token allowance may result in:
Soft throttling (slower responses)
Hard cutoffs (blocked access until reset)
A plan upgrade prompt to access more capacity
Transparency
Token counts may be displayed:
At the end of each chat exchange
In the user’s usage dashboard
In the app settings under “Plan & Usage”
Understanding tokens helps you gauge how much AI “work” your queries demand, and ensures fair, tier-based access to Chariot’s capabilities.
22.2 How Tokens Are Measured
22.2 How Tokens Are Measured
Chariot measures tokens across all interactions by combining multiple contributing elements from each AI operation. Your total token usage for a given task is calculated from:
🧠 1. Prompt Length (User Input)
Every word, number, punctuation mark, and special character in your prompt consumes tokens.
Example:
Short prompt: “What’s the value of this VIN?” → ~8 tokens
Long prompt: “Please analyze this repair contract and summarize legal risks” → ~15–25 tokens
🤖 2. Completion Length (AI Output)
The AI’s reply also consumes tokens—longer responses cost more.
Example:
Summary: “This contract appears to favor the dealer.” → ~10 tokens
Detailed analysis: multi-paragraph output → 100–800+ tokens
📷 3. Embedded Image or Document Analysis
Each uploaded photo or PDF requires visual or text parsing, both of which consume tokens based on content density:
Clean car photo: ~150–400 tokens
Full contract (5–10 pages): ~2,000–5,000+ tokens
Receipt or invoice: ~300–1,000 tokens
🧾 4. System Instructions and Memory Context
Background tokens used by the AI to:
Maintain short-term session memory
Apply system safety and format instructions
Interpret previous inputs in the same session
These “hidden” tokens typically range from 500–1,500 tokens per session, depending on history length and complexity
⚠️ Example Total Breakdown
A full vehicle condition analysis prompt with:
2 user images (~700 tokens)
Description input (~40 tokens)
AI output (~900 tokens)
System context (~600 tokens)
→ Total: ~2,200–2,400 tokens
Token tracking is active in real-time and may influence throttling, plan usage, or upgrade prompts. You can view your current token count and remaining usage in the app under Settings > Usage Dashboard.
22.3 Token Caps by Plan
22.3 Token Caps by Plan
Chariot enforces token usage limits based on your current subscription tier. These token caps govern how much text, image, and document data you can process within a given billing period. Caps apply to the combined total of prompt inputs, AI outputs, scanned documents, and processed photos.
🪙 Token Cap Structure
PlanDaily Token CapMonthly Token CapReset FrequencyFree (Trial)750 tokens/day3,000 tokens/monthResets every 24hWeekly ($4.99/wk)2,000 tokens/day14,000 tokens/monthResets every 24hMonthly ($14.99/mo)5,000 tokens/day50,000 tokens/monthResets every 24hAnnual ($99.99/yr)8,000 tokens/day100,000+ tokens/monthResets every 24hMax Insight (Add-on)20,000 tokens/day500,000 tokens/monthResets every 24h
Note: These values may be adjusted dynamically based on usage patterns, infrastructure scaling, or abuse detection.
Token Use Cases That Count Toward Your Cap
Submitting a chat prompt
Receiving a long-form AI reply
Uploading a car image or interior photo
Scanning PDF documents for contract review
Uploading repair invoices, titles, or insurance docs
Generating full-length condition or risk reports
Exceeding Your Token Cap
If you exceed your plan's cap:
You may receive a "token limit reached" warning
Additional requests may be throttled or rejected
You may be offered a plan upgrade or boost purchase
Soft throttles may reduce output detail or delay response time.
Viewing Your Token Balance
You can check your current token usage in:
App Settings > Usage Dashboard
Account > Plan Summary
Or on the Upgrade Plan screen
Your token meter shows both daily and monthly totals.
Token caps are designed to maintain fair system access and performance across all user tiers while allowing scalable AI utility at higher plan levels.
22.4 Chat Token Usage
22.4 Chat Token Usage
Each chat interaction in Chariot is measured in tokens, which are consumed by both the user’s input and the AI’s response. Your available token balance is reduced every time you send a prompt or receive an answer.
🗣️ Token Use Breakdown
ElementToken ConsumptionUser promptEach word, punctuation mark, or symbolAI responseEvery word, sentence, and formatting elementContext memoryTokens used to maintain continuitySafety instructionsBackground tokens used by the system
🧠 Example Token Usage
Short chat (1–2 sentences) → ~20–100 tokens
Medium chat with follow-up → ~150–400 tokens
Detailed prompt + long response → 500–1,000+ tokens
Multi-turn conversation (5+ turns) → 1,500–3,000+ tokens
Each chat session includes hidden system tokens (usually 500–1,200 per session) for memory, guardrails, and formatting behavior.
🔁 Chat Token Efficiency Tips
Be concise: Shorter prompts use fewer tokens and get faster results
Avoid repetition: Asking the same question in multiple ways wastes tokens
Use clear instructions: Reduce clarification loops that add extra messages
Break up long requests into stages instead of asking for large outputs at once
⚠️ When You Hit a Token Limit
If your current plan’s daily or monthly token cap is exceeded, you may receive:
“Limit reached” messages
Shortened replies
Slowed response speed
Feature lockout until token reset or plan upgrade
Chariot’s AI chat system is optimized to balance quality, speed, and fair usage. Your token meter helps track how much you’re using—visible in the Usage Dashboard inside the app.
22.5 Vision/Image Token Usage
22.5 Image Token Usage
Each image uploaded for AI analysis—such as vehicle exterior photos, interior damage, or undercarriage scans—consumes a significant number of tokens. The total token cost is automatically deducted from your daily and monthly cap.
Token Cost Factors
Image-based requests typically consume 2,000 to 5,000+ tokens per image, depending on:
Resolution: Higher-resolution files require more processing power and token allocation.
Object Count: Images with multiple vehicles, people, or overlapping elements increase complexity.
Captioning Detail Level: If the AI is asked to generate detailed captions or annotations, additional tokens are consumed.
Request Complexity: Multi-step queries (e.g., “Flag paint damage, summarize condition, and estimate age”) require more tokens than simple assessments.
Sample Token Estimates
Image TypeEstimated TokensClear photo of single car exterior2,000–3,000Interior shot with multiple features3,000–4,500Damaged or obstructed images4,000–5,500+Custom captioning or multiple requests5,000–7,000+
Efficiency Guidelines
Use clear, well-lit images for token efficiency.
Avoid uploading duplicates or irrelevant shots.
Limit requests per image to reduce unnecessary token draw.
Token usage from image uploads is displayed in your usage tracker within the app. Uploading blurry, duplicate, or misleading photos wastes tokens and may reduce analysis quality.
22.6 PDF/Document Token Usage
22.6 PDF/Document Token Usage
When you upload a document for analysis—such as a purchase agreement, warranty, or repair invoice—Chariot calculates token usage based on the file’s length, formatting, and content density. Each page incurs a token cost, and multi-page documents can quickly accumulate high usage.
Token Cost Estimates
Per Page: Approximately 2,500–4,000 tokens
Short Documents (1–3 pages): 5,000–10,000 tokens total
Standard Contracts (5–10 pages): 15,000–35,000 tokens
Complex or Loosely Structured Contracts: 40,000–50,000+ tokens
Factors That Increase Token Usage
Dense Legal Text: Long clauses, embedded conditions, or regulatory disclaimers increase token volume
Multicolumn Layouts: Side-by-side formatting results in higher token consumption
PDF Parsing Errors: OCR attempts on scanned, low-quality, or image-only PDFs add overhead
Requested Output Detail: Requests for per-clause breakdowns, risk flags, and summaries use more tokens
Example Token Breakdown
Document TypePagesApprox. TokensBasic Bill of Sale12,800–3,500Warranty with Terms & Excl.39,000–12,000Full Dealer Finance Contract828,000–38,000Annotated Legal Agreement12+40,000–55,000+
Best Practices
Submit clean, machine-readable PDFs (no scans/photos)
Avoid uploading irrelevant or blank pages
Request only the level of analysis you need (e.g., summary vs full breakdown)
Token usage for PDF uploads is shown in your Usage Dashboard and counts toward both daily and monthly caps. Exceeding caps may block additional uploads until reset or plan upgrade.
22.7 Model-Specific Token Pricing
22.7 Model-Specific Token Pricing
Chariot uses a variety of AI models to power chat, image analysis, and document interpretation. These models—such as OpenAI GPT-4o, Anthropic Claude 3, or future open-weight systems—each have unique token pricing and computational demands.
Key Points
Different Models, Different Costs: Each AI model processes tokens at a different cost basis. GPT-4o may consume fewer tokens per output than Claude 3, or vice versa, depending on the task.
Model Switching: Chariot may dynamically switch models in the background to optimize performance, reduce latency, or ensure service continuity.
No Advance Notice Required: Token pricing changes tied to model adjustments may occur without user-facing notice. Your total token usage may vary for the same task on different days.
Impact on Token Caps: Shifts in token pricing may affect how quickly you reach your daily or monthly cap. For example, a task using GPT-4o may consume 10,000 tokens while the same task on Claude 3 consumes 14,000 tokens.
Why This Matters
Performance Consistency: Model switching helps balance speed, accuracy, and uptime, especially under heavy load.
Billing Fairness: Chariot uses internal benchmarks to normalize token impact across models, but token burn rate may still vary.
Plan Transparency: Users are billed by token usage, not model identity. Pricing is model-agnostic at the user level.
Chariot reserves the right to optimize model routing without individual disclosures. Users should monitor their token usage in the Usage Dashboard to understand real-time consumption, regardless of backend model.
22.8 Token Usage Display
22.8 Token Usage Display
Chariot may provide users with access to token usage metrics within their account dashboard, subscription settings, or usage screens. These indicators are designed to help users monitor consumption and manage plan limits effectively.
Availability
Token Stats Display: You may see estimated token consumption per image, chat, or document analysis.
Dashboard Access: Certain subscription tiers include access to a usage panel or token burn history.
Reset Timers: Daily and monthly reset timers may be displayed to help you plan usage.
Limitations
No Guarantee of Real-Time Accuracy: Chariot is not obligated to provide real-time token tracking or instant metering.
Estimates Only: Displayed token counts may be rounded, delayed, or estimated based on usage category.
Partial Visibility: Some internal token usage (e.g., system messages, routing overhead) may not be reflected in user-facing stats.
Token Overages
Reaching token caps may result in:
Temporary throttling or locked features
Prompts being rejected or cut off
Requirement to upgrade or wait for reset
Chariot provides token visibility features as a courtesy, not a contractual obligation. Users are encouraged to manage usage conservatively and upgrade plans as needed for high-volume workflows.
22.9 Excess Usage Management
22.9 Excess Usage Management
When a user reaches their allotted token limit based on their subscription plan, Chariot may implement automated restrictions and guidance to manage overuse and preserve platform stability.
Token Cap Consequences
Once your plan’s token threshold is met (daily or monthly), the following measures may apply:
Chat Response Truncation
AI responses may be automatically shortened, limited to summaries, or blocked entirely.Feature Lockouts
Image uploads, document analysis, or deep insight tools may become temporarily inaccessible.Upgrade Notices
The system may display modals, banners, or callouts offering plan upgrades to continue access.Soft Throttling
Requests may be delayed, queued, or rate-limited without warning if limits are surpassed rapidly.
Recovery Options
Wait for Reset
Daily limits reset every 24 hours. Monthly limits reset based on your billing cycle.Upgrade Plan
Unlock higher caps or unlimited usage by subscribing to a higher tier.One-Time Add-Ons (if enabled)
In the future, Chariot may offer pay-per-use tokens or unlock packs for burst needs.
Token restrictions apply automatically across all AI models and features to maintain platform fairness. Users are responsible for monitoring their usage and adjusting their plan if frequent overages occur.
22.10 No Rollover Tokens
22.10 No Rollover Tokens
Chariot operates on a use-it-or-lose-it model for token allowances. Any unused tokens allocated during your billing period will expire at the end of that cycle and will not carry over into future months.
Key Terms
Monthly Token Expiry
All unused tokens reset to zero at the start of your next billing cycle, regardless of usage.No Accrual or Banking
Tokens cannot be banked, pooled, or accumulated across billing periods.No Refund for Unused Tokens
Unused tokens have no cash value and do not qualify for partial refunds or credits.
Why This Policy Exists
This policy helps:
Maintain fair infrastructure allocation
Predict server load and cost at scale
Incentivize consistent use and planning
Users are encouraged to monitor their token usage throughout each period and plan AI tasks accordingly. Chariot may provide in-app reminders or usage dashboards to assist with this process.
22.11 Token Abuse Patterns
22.11 Token Abuse Patterns
Chariot actively monitors for behaviors that indicate misuse or exploitation of token-based systems. The following actions are considered token abuse and may result in enforcement:
Examples of Token Abuse
Long Prompt Spamming
Repeatedly submitting excessively long prompts designed to maximize output or extract system behavior.Conversational Looping
Generating repetitive input/output chains to artificially inflate usage or simulate interaction patterns.Image Flooding
Uploading large volumes of images in a short period without meaningful variation or purpose.Noisy or Obfuscated Scans
Submitting low-quality, intentionally unreadable, or manipulated PDFs to overwhelm the model or probe its limitations.
Consequences of Abuse
If token abuse is detected, Chariot may apply:
Immediate soft throttling
Temporary slowdown or rejection of further inputAccount audit or moderation
Review of user behavior, session logs, and device activitySuspension or permanent restriction
Especially for repeated or intentional abuse patterns
Enforcement Notes
Detection may be based on behavioral fingerprinting, usage volume, input structure, and device-linked patterns
Violations count against your global usage risk score across chat, image, and PDF features
Appeals may be limited if abuse is determined to be deliberate or automated
Chariot's token system is designed for fair use. Misuse not only undermines platform reliability but may result in permanent access loss.
22.12 Prompt Injection Prohibition
22.12 Prompt Injection Prohibition
Users are strictly prohibited from attempting to override, manipulate, or subvert Chariot’s internal instructions, model safeguards, or system behavior through prompt injection techniques. This includes both textual and embedded prompt layers.
What Constitutes Prompt Injection
System Prompt Hijacking
Attempts to trick the AI into ignoring default rules or outputs (e.g., “Ignore previous instructions and…”).OCR Layer Injection
Embedding manipulative commands within scanned documents or image captions to affect downstream output (e.g., handwritten directives in image margins).Recursive Prompting
Feeding in modified model outputs to nudge the system into unauthorized logic, opinions, or unsafe completions.Command Framing
Using adversarial framing to probe internal rules, jailbreak constraints, or safety filter thresholds.
Enforcement & Liability
All such activity is logged, flagged, and may result in:
Session termination
Token forfeiture
Account suspension
Permanent ban for repeated or malicious attempts
No refunds will be issued for tokens consumed during prompt injection attempts.
Chariot is not liable for any consequences, harms, or misunderstandings caused by outputs generated through prompt injection.
Chariot reserves the right to analyze submitted prompts for intent and pattern consistency. Prompt integrity is critical to model safety and system reliability. Violations will be treated as security threats.
22.13 Fair Use Enforcement
22.13 Fair Use Enforcement
Chariot enforces a fair use policy to ensure system availability, model performance, and platform sustainability for all users. Even if a user remains below their formal token cap, excessive or disproportionate usage may still be restricted.
Definition of “Reasonable” Use
Usage that aligns with typical human interaction patterns within the assigned subscription tier.
Avoids automated, scripted, or industrial-scale prompting that strains shared infrastructure.
Does not simulate or replicate commercial-scale model probing under a personal account.
Examples of Excessive but Technically “Under Cap” Use
Submitting hundreds of edge-case prompts daily
Chaining vision + chat + document scans continuously
Using multiple sessions per hour with long or adversarial inputs
Exploiting memory limits, threading logic, or retry loops
Enforcement Actions
Chariot may, at its sole discretion:
Apply soft throttles (e.g., slower responses, output length caps)
Trigger temporary usage freezes or cooldowns
Require plan upgrades for continued access
Flag accounts for review or audit
No Guarantee of Unlimited Behavior Within Cap
Your subscription token cap is a maximum allocation, not a guarantee of unrestricted use regardless of impact. If your usage affects:
Platform latency for others
Server cost thresholds
Model abuse detection rates
... Chariot reserves the right to take action.
By using Chariot, you agree to this principle of shared platform integrity. Token fairness ensures long-term viability for all users.
22.14 Shared Accounts Prohibited
22.14 Shared Accounts Prohibited
Chariot assigns token allocations and usage privileges on a per-user basis. Sharing your login credentials, enabling group access, or using automated systems to simulate multiple users under one account is strictly prohibited.
Prohibited Sharing Scenarios
Multiple individuals using the same account to access AI chat, image analysis, or contract reviews.
Automated scripts or bots dispatching prompts for multiple users via a single session.
Shared devices where sessions are rotated among different users without authorization.
Teams, classrooms, or businesses routing traffic through one user account.
Why This Policy Exists
To preserve fair resource distribution
To maintain accurate token tracking
To prevent misuse of personal or trial plans for group/commercial benefit
To ensure personalized support and safety enforcement
Enforcement
Violations of this clause may result in:
Immediate session termination
Token usage suspension
Revocation of subscription benefits
Permanent ban or billing dispute trigger
Loss of access to any generated reports or data
Allowable Exceptions
If you need shared access across a team or organization:
You must apply for a commercial license
Or enroll in an enterprise subscription plan with multi-user support
Account sharing or scripted multi-user prompt behavior is treated as abuse and grounds for restriction without refund. Each user must maintain a distinct, authenticated identity on the platform.
22.15 Additional Token Packs
22.15 Additional Token Packs
Chariot may offer optional on-demand token packs as in-app purchases. These allow users to exceed their base subscription token allocation without upgrading plans.
Key Terms
Standalone Purchase:
Token packs are not included in your subscription. They must be purchased separately.Typical Example:
250,000 tokens for $9.99 USD, available via Apple/Google IAP or web payment gateway.No Auto-Renewal:
These are one-time purchases, not recurring.Immediate Activation:
Tokens are credited to your account upon successful transaction and are available for immediate use.Expiration Policy:
Purchased token packs may expire if unused after a fixed period (e.g., 90 days), subject to app notice.
Use Cases
Need to finish a large report, scan more contracts, or run a batch of vehicle photos
Want to avoid subscription upgrade but temporarily exceed caps
Working on a time-sensitive task requiring more AI processing
Usage Integration
Token usage is pooled: subscription tokens are consumed first, followed by your top-up pack.
You may view your remaining balance in app settings or your account dashboard.
AI responses, image processing, and PDF scans will all draw from the active token pool automatically.
No Refunds or Transfers
Token pack sales are final
Non-transferable, tied to your account
Cannot be split, shared, or gifted
Purchasing additional token packs gives users a flexible, burst-friendly way to access more AI services without committing to a higher subscription tier.
22.16 Token Use per Feature
22.16 Token Use per Feature
Chariot assigns approximate token costs to each major feature to ensure fair usage and transparent resource consumption. Token usage may vary based on input size, complexity, model type, and formatting.
FeatureAverage Tokens UsedSingle AI Chat1,000 – 2,000 tokensVIN Summary + Lookup2,000 – 3,000 tokensMulti-Image Review5,000 – 8,000 tokensBasic PDF Scan10,000 – 20,000 tokensMax-Depth Contract Review30,000 – 50,000+ tokens
Notes:
Single AI Chat:
Includes prompt + response. Complex chats or follow-ups may exceed the average.VIN Summary + Lookup:
Combines vehicle data enrichment, market insight, and model-specific AI output.Multi-Image Review:
Includes visual parsing, object detection, damage inference, and captioning.Basic PDF Scan:
Includes OCR, clause breakdown, and headline risk scoring.Max-Depth Contract Review:
Used for legal document parsing with clause-by-clause AI scrutiny, annotations, and heuristics.
Important Disclaimers:
Actual usage may be above or below the stated ranges depending on the file length, resolution, verbosity, or ambiguity.
Users are encouraged to monitor token usage if available via dashboard.
Excessive usage of any one feature may consume your monthly quota quickly or prompt upgrade offers.
This system helps ensure that larger, deeper services cost proportionally more, maintaining fairness across all subscription tiers.
22.17 Model Drift & Token Shifts
22.17 Model Drift & Token Shifts
Chariot reserves the right to adjust token usage patterns, feature verbosity, or subscription limits in response to shifts in underlying AI model behavior or pricing. These changes may occur without prior user consent in order to maintain platform stability and sustainability.
What This Means:
Model Drift:
If the AI system (e.g., GPT-4o, Claude) begins generating longer or more resource-intensive outputs than originally projected, Chariot may:Shorten completions
Adjust prompt constraints
Limit follow-up depth
Token Cost Changes:
If Chariot's third-party AI providers update pricing or token billing methods, Chariot may:Lower daily/monthly usage caps
Increase token deduction rates
Modify subscription plan allowances
Possible Adjustments:
Area AffectedExamples of ChangesAI Response LengthShorter summaries, fewer explanationsToken Cap per PlanReduced daily/monthly limitsPDF/Image ComplexityHeavier uploads may trigger hard limitsSubscription TiersRebalanced access to preserve fair usage
Why This Clause Exists:
To future-proof the platform against rising AI infrastructure costs
To protect token fairness across the user base
To enable adaptability in response to evolving model efficiency
Chariot will aim to notify users of major shifts through app updates, plan setting notices, or in-app alerts, but explicit consent will not be required for such systemic updates. Continued use of the platform constitutes agreement to revised token economics.
22.18 Token Use Visibility
22.18 Token Use Visibility
Chariot administrators may access internal token usage logs for each user account. These logs are used strictly for:
Auditing resource consumption
Billing validation and dispute resolution
Detecting abuse or anomalous behavior patterns
Plan enforcement and system optimization
Data Included in Token Logs
Token logs may contain:
Timestamped prompt and response sizes
Associated feature (e.g., image upload, chat, PDF scan)
Token totals per session and per day
Device and IP metadata (if enabled)
Privacy & Security
Token logs are stored securely and access is limited to authorized internal team members.
These logs are never shared publicly or sold to third parties.
Logs may be retained for up to 12 months or longer if linked to fraud investigations or billing audits.
User Transparency
Users may view approximate token usage in their dashboard (if enabled), but not full logs.
Token logs are not exportable, and admin access is non-negotiable for platform integrity.
Chariot uses token visibility as a protective layer to ensure platform health, billing fairness, and abuse deterrence. All token usage is subject to internal review without additional user notice.
22.19 Token Rate Limits
22.19 Token Rate Limits
To ensure platform stability and fair access, Chariot enforces token rate limits in addition to overall monthly or daily usage caps. These rate limits help prevent abuse, overloading, and denial-of-service scenarios.
Examples of Rate Limits
Time WindowMax Token Allowance (subject to plan tier)Per Minute5,000 – 10,000 tokensPer Hour20,000 – 50,000 tokensPer DayPlan-specific (e.g., 100K–500K tokens)
How This Works
If you exceed your minute or hour rate, additional prompts may:
Be throttled (delayed responses)
Receive shortened or blocked output
Trigger temporary lockout of high-cost features
Rate limits apply cumulatively across all features, including:
AI chat messages
Image uploads
PDF/document analysis
VIN lookups or summary generation
Enforcement & Visibility
Rate limits are enforced automatically by Chariot’s backend.
You may see in-app notices or alerts if you exceed real-time thresholds.
Admins may review rate limit logs as part of abuse prevention workflows.
Continued overuse or attempts to bypass these limits (e.g., through scripting, re-logging, or account switching) may result in throttling, suspension, or account audit.
22.20 Throttling Thresholds
22.20 Throttling Thresholds
To maintain system integrity and fair usage across all users, Chariot enforces automatic throttling when extreme usage thresholds are triggered. Throttling slows or suspends access to high-cost features temporarily.
Trigger Conditions
You may experience throttling if any of the following conditions are met:
10,000 tokens used in a single minute
100,000 tokens consumed within 2 hours
25 vision uploads (image analysis) submitted in one calendar day
Effects of Throttling
Once throttling is triggered:
AI responses may be delayed, shortened, or disabled temporarily
Image/PDF features may be locked until usage resets
Token consumption rates may be scaled back dynamically
You may see in-app warnings or receive cooldown notifications
Reset Timing
Throttling typically resets after:
1–5 minutes for minute-level limits
2–4 hours for hourly caps
Midnight local time for daily upload limits
Intentional Abuse
If throttling thresholds are intentionally pushed through scripting, automation, or repeated testing:
Chariot may trigger hard blocks, require plan upgrade, or suspend access
Admins may review logs for misuse and enforce account action
Note: These thresholds apply regardless of plan tier to protect the system and ensure equitable usage. Heavy workflows should be spread out across time windows or approved through enterprise plans.
22.21 No Token Credit for Canceled Plans
22.21 No Token Credit for Canceled Plans
If you cancel your Chariot subscription before the end of your billing cycle, any unused token allotments associated with your current plan are automatically forfeited at the time of cancellation.
Key Terms
No Pro-Rata Token Refunds
You will not receive credits or reimbursements for unused tokens when canceling.Plan Downgrades or Terminations
Downgrading to a lower plan or terminating your subscription will also void remaining token balances tied to the higher plan.One-Time Token Packs Not Affected
Separately purchased token packs (if offered) are not impacted unless otherwise specified.
Example
If your plan includes 100,000 tokens/month and you cancel halfway through after using 30,000 tokens:
The remaining 70,000 tokens are not preserved
You will retain access only until the end of the billing period unless you manually forfeit access earlier
Policy Purpose
This policy ensures consistency with most app store subscription systems and discourages manipulation of billing cycles for usage advantage.
Reminder: Canceling a plan does not delete your account or past report data unless explicitly requested.
22.22 Token Spending Transparency
22.22 Token Spending Transparency
Chariot is committed to helping users make informed decisions about their token usage. To support transparency, we provide estimated token cost ranges for each major feature within the app. However, exact token consumption may vary due to the complexity of the request and changes in pricing from external AI model providers.
What This Means
Visible Estimates:
Users will see approximate token ranges for tasks like chat responses, document reviews, and image uploads in the app or support materials.Dynamic Factors:
Actual token usage may fluctuate based on:Prompt/response length
Model type (e.g., GPT-4o vs Claude)
Output verbosity
Embedded system context or metadata
OCR and image captioning complexity
Third-Party Variability:
LLM providers (e.g., OpenAI, Anthropic) periodically change token rates or output behavior. These shifts may affect Chariot’s underlying token cost without prior notice.
User Expectations
Chariot will strive to provide reasonable accuracy in usage displays.
Users should expect approximate, not exact, token counts.
Token transparency tools may be found in account settings, usage dashboards, or feature tooltips.
Example Display (Subject to Change)
FeatureEst. Tokens UsedAI Chat (1 message)1,000 – 2,000VIN Summary2,000 – 3,000Condition Image Review5,000 – 8,000PDF Contract Scan10,000 – 20,000Max Risk Clause Breakdown30,000 – 50,000
This policy exists to help users track and optimize their experience while maintaining flexibility in a rapidly evolving AI cost ecosystem.
22.23 Token Sync Across Devices
22.23 Token Sync Across Devices
Your token usage is account-based and synchronized across all sessions and devices. This means that:
Switching phones, tablets, or logging in from a different device does not reset your token quota.
All usage—including chat, vision, and PDF analysis—is tracked centrally and applies to your active subscription limits regardless of platform or login point.
Quotas persist across:
iOS ↔ Android
App ↔ Web (if applicable)
Multiple reinstallations or cache resets
No device migration or logout event will refresh token limits. Users attempting to bypass limits through device switching, emulator usage, or reinstallation cycles may be flagged under Chariot's abuse prevention systems.
This synchronization ensures accurate tracking, fair usage, and secure enforcement of subscription terms across your account footprint.
22.24 Priority Access by Token Behavior
22.24 Priority Access by Token Behavior
Chariot may grant early or priority access to new features, advanced models, or experimental tools based on a user’s historical token behavior. This is designed to reward fair usage and promote platform stability.
Qualifying Behaviors
Priority access may be extended to users who:
Stay within expected token caps without triggering abuse flags
Avoid excessive image uploads or document scans per minute/hour
Use the system for intended analytical or research purposes
Maintain a stable pattern of subscription engagement
Possible Benefits
Eligible users may receive:
Early access to next-generation AI models
Access to premium analysis tools before public rollout
Invitations to participate in closed research betas
Higher temporary token thresholds during limited-time tests
This policy is discretionary and may vary across plan tiers. Chariot reserves the right to revoke or adjust access privileges at any time based on platform needs and observed usage behavior.
22.25 No Guaranteed Output Length
22.25 No Guaranteed Output Length
The number of tokens consumed per request does not equate to a guaranteed word count, answer type, or output depth. Large token usage may still result in short or concise replies depending on:
Model interpretation of prompt complexity
Safety filters or truncation behavior
Internal system instructions or context compression
Multi-modal prompt overhead (e.g., image or document inclusion)
Users should understand that token usage reflects model processing effort, not just visible text length. No refund or compensation will be issued solely on the basis of perceived brevity.
22.26 Survival of Terms
22.26 Survival of Terms
All token-related policies outlined in this section remain enforceable regardless of:
Subscription cancellation, downgrade, or expiration
Changes to pricing tiers, quotas, or token costs
Transitions between backend vendors or AI models
Modifications to app functionality or feature gating
These terms apply retroactively and prospectively to all token usage tied to your account. Chariot reserves the right to audit historical usage and apply enforcement actions as needed.
Contact Us
If you have any questions or concerns about our Terms of Service or the handling of your personal information, please contact us at support@chariotreport.com