Kasava
Back to Blog
Product

Transparent Pricing: How We Calculate AI Credits

Ben Gregory-Jan 24, 2026

Why We're Writing This

At Kasava, we believe pricing should be transparent. When you use AI features, you should know exactly what you're paying for and why. Here's exactly how we calculate credits, what they cost us, and what we charge you.

The Credit System

Basic Conversion

Our pricing is simple:

300 credits = $1 USD

This rate is calculated to maintain approximately 11% margin above what we pay Anthropic for Claude API access. That margin covers our infrastructure, token tracking, caching, and operational costs.

Token to Credit Conversion

AI models charge differently for input (what you send) vs. output (what the model generates). Output tokens are more expensive because they require more computation.

Token TypeCredit CostRationale
Input tokens1 credit per 1,000 tokensMirrors Anthropic's 1:1 base pricing
Output tokens5 credits per 1,000 tokensReflects Anthropic's 5x output pricing

Real Example:

A typical chat interaction might use:

  • 5,200 input tokens (your question + context)
  • 1,500 output tokens (Claude's response)

Credit calculation:

  • Input: floor(5,200 ÷ 1,000) × 1 = 5 credits
  • Output: floor(1,500 ÷ 1,000) × 5 = 5 credits
  • Total: 10 credits

User-Favorable Rounding

Notice we use floor() (round down), not ceil() (round up). This means:

  • 999 input tokens = 0 credits (not 1)
  • 1,001 input tokens = 1 credit (not 2)
  • Only whole credits are ever deducted

This costs us slightly more, but it's the right thing to do. You shouldn't pay for tokens you didn't use.

The Math Behind Our Margin

Let's be completely transparent about our costs and margins.

What We Pay Anthropic

Claude Sonnet 4.5 (our primary model):

  • Input: $3.00 per million tokens
  • Output: $15.00 per million tokens

What We Charge You

At 300 credits per dollar:

  • Input: 1,000 tokens = 1 credit = $0.00333
  • Output: 1,000 tokens = 5 credits = $0.01667

The Margin Math

For every $1 you spend on credits:

UsageOur RevenueOur CostMargin
300K input tokens$1.00$0.9011.1%
60K output tokens$1.00$0.9011.1%
Mixed (typical)$1.00~$0.90~11%

We've optimized for user value over maximum extraction. If we can lower it in the future, we will.

Plan Allocations

Each plan includes monthly credits per seat:

PlanCredits/Seat/MonthDollar ValuePrice/Seat
Free300~$1$0
Professional3,000~$10$35
EnterpriseUnlimitedCustomContact us

Free users get enough credits for approximately 300,000 input tokens or 60,000 output tokens per month—plenty for evaluation and light usage.

Credit Refill Packages

Need more credits? We offer volume bonuses:

PurchaseCreditsEffective RateBonus
$103,000300/$Base rate
$258,000320/$+7% bonus
$5016,000320/$+7% bonus
$10035,000350/$+17% bonus

Larger purchases get better rates because they reduce our payment processing overhead.

Auto-Refill

Pro and Enterprise plans can enable auto-refill:

  1. Set a threshold (e.g., 500 credits)
  2. Choose a refill amount ($10, $25, $50, or $100)
  3. When balance drops below threshold, we automatically purchase more

You stay in control with monthly spending caps. Free plans cannot enable auto-refill to prevent unexpected charges.

What We Track

For billing accuracy and your visibility, we track input and output tokens separately for each AI feature:

  • Chat: Conversations with your codebase
  • Plan Generation: PRD and spec creation
  • Bug Analysis: Chrome extension analysis
  • Document Processing: Google Docs AI
  • Code Analysis: Repository intelligence
  • Commit Analysis: Semantic commit insights

Repository Indexing (Quota-Based)

Repository indexing is tracked separately from token-based credits. Full repository indexing is an expensive operation involving GitHub API calls, code parsing, and embedding generation.

PlanFull Indexes/MonthOverage Cost
Free56 credits each
Professional506 credits each
EnterpriseUnlimitedN/A

What counts as an index?

  • Full repository index or reindex: 1 operation
  • Incremental updates from pushes: Free (not counted)
  • Selective file reindexing: Uses Code Analysis quota instead

You can see your usage breakdown in Settings → Billing.

Why This Approach?

Credits vs. Direct Token Billing

We considered billing tokens directly (like OpenAI's API), but credits offer advantages:

  1. Predictable costs: Round numbers are easier to budget
  2. Future flexibility: We can add models without changing your billing
  3. Fair aggregation: Mix of input/output is handled transparently
  4. User-favorable rounding: Small operations cost nothing

Why ~11% Margin?

We chose a slim margin for several reasons:

  1. Trust building: Transparent pricing builds long-term relationships
  2. Market position: We compete on value, not margin extraction
  3. Sustainability: 11% covers costs while remaining accessible
  4. Alignment: Our success depends on you using more, not paying more per use

Questions?

We're committed to pricing transparency. If anything about our billing is unclear, reach out:

  • Email: support@kasava.dev
  • Dashboard: Settings → Billing shows real-time usage
  • Documentation: docs.kasava.dev/billing

This pricing is effective January 2026. We'll update this post if our pricing model changes.