Enterprise RAG Chatbot API Cost Calculator (Prompt Caching Enabled)
Estimate monthly LLM API bills for enterprise Retrieval-Augmented Generation (RAG) chatbots with long retrieved context windows and 80% prompt caching.
Describe Your Workload
= 75,000 requests a month (30 days)
System prompt + chat history + retrieved documents + the user's message.
Count too: they bill as output.
Rule of thumb for English: is roughly 750 words.
With , the repeated start of each prompt is billed at the cheaper cache-read rate.
GPT-6.1 Sol
- Input: 60M tokens @ $2/M
- $120.00
- Cached input: 240M tokens @ $0.1/M
- $24.00
- Output: 30M tokens @ $10/M
- $300.00
- Total per month
- $444.00
Cost = tokens ÷ 1,000,000 × price per 1M, summed for input, cached input and output
Cheapest mid-tier model for this workload: Mistral Large 4 at $121.80/month (73% less). Cheapest overall: DeepSeek V4.1 Flash at $22.02/month. Cheaper models can be weaker, so test them on your own prompts before switching.
This Workload on Every Model
300M input and 30M output tokens a month at real-time prices. Cheapest first; tap a model to load it above.
| Model | Tier | Input / 1M | Cached / 1M | Output / 1M | Context | Monthly cost | vs GPT-6.1 Sol |
|---|---|---|---|---|---|---|---|
| DeepSeek | Budget | $0.045 | $0.018 | $0.5 | 1M | $22.02 | -95% |
| Anthropic | Budget | $0.1 | $0.01 | $0.5 | 1M | $23.40 | -95% |
| OpenAI | Budget | $0.1 | $0.01 | $0.5 | 1.1M | $23.40 | -95% |
| Alibaba Qwen | Budget | $0.15 | $0.016 | $0.47 | 1M | $26.94 | -94% |
| Budget | $0.25 | $0.025 | $1.5 | 1M | $66.00 | -85% | |
| Meta (open weights) | Budget | $0.35 | $0.17 | $1 | 524.3K | $91.80 | -79% |
| Mistral | Mid | $0.7 | $0.07 | $2.1 | 1M | $121.80 | -73% |
| DeepSeek | Mid | $0.103 | $0.098 | $4.11 | 1M | $153.00 | -66% |
| Mid | $0.825 | $0.0825 | $4.125 | 1M | $193.05 | -57% | |
| OpenAI | Mid | $1 | $0.05 | $5 | 1.1M | $222.00 | -50% |
| xAI | Flagship | $2 | $0.5 | $6 | 500K | $420.00 | -5% |
| Anthropic | Mid | $2 | $0.1 | $10 | 1M | $444.00 | 0% |
| Flagship | $2 | $0.1 | $10 | 1M | $444.00 | 0% | |
| OpenAI | Mid | $2 | $0.1 | $10 | 1.1M | $444.00 | selected |
| Anthropic | Flagship | $4 | $0.2 | $20 | 1M | $888.00 | +100% |
| Anthropic | Flagship | $10 | $0.25 | $50 | 1M | $2,160.00 | +386% |
| OpenAI | Flagship | $10 | $1 | $50 | 1.1M | $2,340.00 | +427% |
Prices are the per-token rates OpenRouter listed on 5 Nov 2026. Buying direct from a provider can be priced differently, and some providers charge more for very long prompts, so check the provider's pricing page before you commit a budget. Cache-write charges (paid the first time a prefix is cached) are not included; with a high cache share they add little. A month is 30 days.
Notes for This Preset
An enterprise RAG chatbot serving 2,500 questions/day (75,000 requests/month) with 4,000 retrieved context tokens and 400 response tokens costs $444.00/month ($5.92 per 1,000 requests) on GPT-6.1 Sol with 80% prompt caching (5 November 2026). Without prompt caching, reading 300 million input tokens would cost $600.00, pushing the total monthly bill to $900.00. Caching the shared system instructions and retrieved knowledge chunks cuts monthly expenditure by $456.00 (51%). On DeepSeek V4 Pro, the same workload costs $153.00/month.
How can I lower the API bill for this workload?
Three levers, in order of effort: cache the repeated start of the prompt (system prompt, documents, tool definitions), move work that can wait to the Batch API, and test a cheaper model on your own prompts. Shorter answers help too, because output tokens cost several times more than input tokens.
Building an AI Chatbot, Agent or RAG App?
Vsurge Media builds production AI assistants, RAG search and automation for businesses, and helps keep the monthly API bill under control.