Cost Tracking

Risicare automatically tracks LLM costs across 14 providers using a built-in per-model pricing table. The LLM Cost KPI card on the main dashboard shows total spend, and the Models chart breaks down requests by model — visible in the dashboard overview.

Automatic Cost Calculation

The SDK captures token counts on every LLM call; Risicare calculates the dollar cost server-side from those tokens using current model pricing. You don't compute cost yourself.

import risicare
from openai import OpenAI
 
risicare.init()
 
client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}]
)
 
# The SDK captures token counts:
# - gen_ai.usage.prompt_tokens: 10
# - gen_ai.usage.completion_tokens: 15
#
# Risicare then computes the cost server-side and stores it as:
# - llm_cost_usd: 0.000175   (stored cost column)

Supported Providers

* The cached rate is part of the pricing model but is not yet applied end-to-end — see Cache Token Support below.

Pricing Examples

Key model pricing (per 1M tokens, as of February 2026):

OpenAI

Anthropic

Google

Cache Token Support

When end-to-end cached-token support lands, the SDK will read prompt_tokens_details.cached_tokens (OpenAI) and cache_read_input_tokens (Anthropic), thread the count through the gateway and a new storage column, and the server will apply the cached rate shown in the pricing tables above. Until then, enabling provider-side prompt caching still saves you money on your provider bill — it just isn't yet reflected in the cost Risicare reports.

# Provider-side caching works as usual and reduces your real provider bill:
response = anthropic.messages.create(
    model="claude-sonnet-4-5-20250929",
    system=[{
        "type": "text",
        "text": long_prompt,
        "cache_control": {"type": "ephemeral"}
    }],
    messages=[...]
)
# Note: the cached-token discount is not yet applied to the llm_cost_usd
# Risicare stores — that figure is computed on full input tokens for now.

Dashboard Views

Cost by Provider

View total cost breakdown by provider:

OpenAI:     $142.50 (45%)
Anthropic:   $98.20 (31%)
Google:      $45.30 (14%)
Others:      $31.00 (10%)

Cost by Model

See which models cost the most:

gpt-4o:                $98.50
claude-sonnet-4-5:     $67.20
gpt-4o-mini:           $22.00
gemini-2.0-pro:        $18.30

Cost by Feature

Track costs per feature or endpoint:

/api/chat:           $145.00
/api/summarize:       $67.00
/api/search:          $32.00

API Access

Cost data is available per-trace and per-span via the standard management API:

# Get traces with cost data
curl "https://app.risicare.ai/api/v1/traces?limit=50" \
  -H "Authorization: Bearer rsk-..."

Each trace includes total_cost_usd and each span includes llm_cost_usd in the response. Use the dashboard for aggregated cost breakdowns by provider, model, or time period.

Cost Alerts

Create an alert rule that fires when daily spend crosses a threshold:

curl -X POST "https://app.risicare.ai/api/v1/alerts" \
  -H "Authorization: Bearer rsk-..." \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Daily LLM spend over $100",
    "metric": "daily_cost_usd",
    "operator": "gt",
    "threshold": 100.0,
    "window_minutes": 1440,
    "severity": "warning"
  }'

name, metric, operator, and threshold are required. Unknown fields are rejected with 422 — the request body is strictly validated.

Supported metrics. A rule naming any other metric is skipped by the evaluator (logged as a warning) and will never fire:

error_rate · agent_failure_rate · latency_p50 · latency_p95 · latency_p99 · trace_volume · daily_cost_usd · total_tokens

Cost Optimization

When the diagnosis pipeline analyzes a budget- or resource-related error, it can surface cost-oriented fix recommendations, such as a "use a cheaper model when approaching budget" suggestion. These appear as recommendations in the diagnosis output — they are not auto-applied (fix deployment is on the roadmap; see Self-Healing).

• Model downgrade suggestions: e.g. "Switch to a cheaper model when approaching the budget limit"
• Caching / token-reduction hints surfaced alongside a diagnosis, where applicable

Risicare does not currently run a standalone, always-on cost-optimization analysis independent of the diagnosis pipeline.

Next Steps