BOLD Rewards — LLM Cost Calculator

Monthly LLM API cost estimate for the EGR / BOLD Rewards platform. Pricing as of May 2026 — sources: Gemini · Claude · OpenAI.

Provider quick-set

Quick-set assigns a sensible default model for every scenario. You can still mix and match below.

Scale preset

Scale

%

Other

%

📄 OCR (svc-ocr) $0

💬 NLQ Analytics (admin chat) $0

🎓 AI Trainer $0

%

🎁 Content Recommendations $0

%

🔍 Embeddings (RAG) $0

M
Total monthly
$0/mo
Yearly
$0/yr
Per tenant
$0/mo
Per MAU
$0/mo

Provider head-to-head comparison

Same workload, same scale, every scenario routed to the cheapest reasonable model of each provider.

Breakdown by scenario and model

Scenario Provider / Model Requests / mo Input tokens (M) Output tokens (M) Cost / mo ($) % of total
Assumptions: Tier — Paid Standard (per-provider direct API; no Bedrock / Vertex partner uplift). Context caching at ~75% hit rate on system prompts & schemas — cache reads are ~10× cheaper than fresh input on all three providers. PDF / image input is counted as ~1300 tokens per page. Retry rate inflates all volumes proportionally. Prices reflect the public docs as of May 2026 — verify before quoting: Gemini, Claude, OpenAI. For embeddings, Anthropic doesn't ship a first-party model — the conventional pairing is Voyage AI (added in the dropdown).

Sensitivity: what drives cost the most

Driver Current value ×2 → new total Δ $ / mo Δ %
If the given driver doubles (all else equal) — how much the monthly cost changes.