Wavo credits and limits

Wavo usage is metered in credits: 1 credit = $0.001. Each message charged at input_cost + output_cost per 1K tokens (per model in chat-models.ts) with MARKUP 1.1× applied. Owner / admin set monthly cap via PUT /api/v1/chat/credits/cap. Top up via Stripe (POST checkout → purchase → confirm 3DS). View at /settings/team/billing.

Who is this for

Anyone wondering "how much does Wavo cost?" or "why am I out of credits?". Admins managing team usage caps.

The pricing model

Per-message cost

Each Wavo message charges based on:

• Input tokens × input_cost_per_1k (model-specific)

• Output tokens × output_cost_per_1k (model-specific)

• MARKUP: 1.1× (10% on top of base provider cost)

Formula:

cost_credits = (input_tokens × input_cost + output_tokens × output_cost) × 1.1

Stored in chat_message.credits_used per message.

Model costs (verified chat-models.ts)

Approximate cost per 1K tokens in credits (input / output):

These are credit values per 1K tokens (1 credit = $0.001). MARKUP 1.1× already factored.

Why some models cost more

• Premium reasoning (Opus, GPT-4.1, Gemini Pro): better for complex analysis + multi-tool chains

• Balanced (Sonnet — DEFAULT, GPT-4.1 mini): everyday use

• Fast + cheap (Haiku, Flash, kimi-k2, deepseek-chat): simple questions, high volume

Pick model per task. For most daily mediabuyer questions: Sonnet (default) is the right balance.

The credit_balance table

Each team has one credit_balance row:

Same model as Creative Hub credits (ch-112) but separate ledger entries for chat messages.

View balance + history

/settings/team/billing shows:

• Current balance (in credits + USD equivalent)

• Monthly usage vs cap

• Recent chat transactions (per-message cost, model, timestamp)

• Top up button (opens Stripe)

GET /api/v1/chat/credits returns balance. GET /api/v1/chat/credits/history returns paginated transactions.

Top up via Stripe

For Stripe-enabled workspaces:

1. /settings/team/billing → Top up

2. Pick amount or custom value

3. POST /api/v1/chat/credits/checkout → Stripe checkout URL

4. Complete payment (3DS confirm if required: POST /confirm)

5. Webhook → credit_transaction credit entry → credit_balance.balance increments

Visible at /settings/team/billing after webhook completes (usually < 1 min).

Monthly cap (admin)

Owner / admin can set a cap to limit per-period usage:

PUT /api/v1/chat/credits/cap body:

• monthly_cap_credits (integer, in credits)

• cap_reset_at (typically first of month)

When monthly_used >= monthly_cap: chat messages return 402 with reason monthly_cap_reached.

Reset: at cap_reset_at timestamp, monthly_used resets to 0.

Per-user cap?

Currently: cap is per-team (owner-level), not per-individual-user. All team members draw from the same pool.

For per-user discipline: monitor usage in chat_message aggregations, or use Wevion roles to restrict who can use Wavo.

Cost-saving tips

Use cheaper models for iteration

• Iterate prompts with Haiku / Flash / GPT-4.1 mini (1-5× cheaper)

• Switch to Opus / GPT-4.1 / Gemini Pro for final / important questions

Per-session model override in chat UI; per-user preference at /settings → AI preferences.

Use project instructions

Setting context once via project instructions avoids re-typing context per session. Less input tokens = less cost.

Be specific in prompts

Vague prompt → Wavo follow-ups → multiple turns. Specific prompt → single turn, lower total cost.

Avoid unnecessary tool calls

If you already know the data, don't make Wavo re-query. Reference previous answer.

Memory facts persist context

Saving "ROAS target = 3 for ClientA" as a project fact means you don't restate it each session.

Limit output length

"In one sentence" or "Just the number" — reduces output tokens.

What charges, what doesn't

Failed messages

If a model call fails (provider outage, network):

• chat_message.status = failed

• credits_used = 0 (no charge for failed completions)

• Wavo says "I couldn't complete that — please try again"

Partial outputs may or may not charge depending on what reached the user (typical: charge for tokens streamed).

Compliance + audit

credit_transaction is append-only:

• Every charge: type + amount (negative) + chat_message_id reference

• Every top-up: type + amount (positive) + stripe_payment_intent_id

• Sort by date for monthly reports

For audit: export from /settings/team/billing → CSV.

Common surprises

• "Used 1000 credits in a day": long sessions with many tool calls + Opus model add up; check usage breakdown

• "Monthly cap hit early in month": usage spiked (large project audit); raise cap or wait reset

• "Credit balance after top-up": Stripe webhook can take 1-2 min; refresh page

• "Failed messages charged me": typically no charge; if charged, check credits_used on the message + contact support if anomalous

Common issues

• 402 monthly_cap_reached: cap hit; admin raises cap or wait reset

• Out of balance: top up via Stripe

• Top up failed: Stripe 3DS required → use POST /confirm flow

• Different cost vs expected: model + length + tool calls all affect cost; review per-message breakdown in history

Related

• Privacy + data usage — what's stored where

• Pricing tiers — credit bundles in your plan

• Creative Hub AI credits — sibling credit ledger