Estimate 2026 enterprise LLM API token costs, prompt caching savings, and developer benchmarks.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Estimate 30-day enterprise LLM API spend and prompt-caching savings across GPT-5, Claude 4.5 Sonnet, Gemini 2.5 Pro, OpenAI o3, DeepSeek V3.2, and Gemini 2.5 Flash.
Model ID: gpt-5, claude-4-5-sonnet, gemini-2-5-pro, openai-o3, deepseek-v3-2, or gemini-2-5-flash
Average input tokens per request (default 8000)
Average output tokens per request (default 800)
Prompt cache hit ratio from 0 to 1 (default 0.75)
Daily API requests (default 10000)
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http llm-finops-suite https://omnitoolcalc.com/mcp
Estimate 30-day enterprise LLM API spend and prompt-caching savings across GPT-5, Claude 4.5 Sonnet, Gemini 2.5 Pro, OpenAI o3, DeepSeek V3.2, and Gemini 2.5 Flash.
| Argument |
|---|
| Type |
|---|
| Description |
|---|
| model* | string | Model ID: gpt-5, claude-4-5-sonnet, gemini-2-5-pro, openai-o3, deepseek-v3-2, or gemini-2-5-flash |
| input_tokens | number | Average input tokens per request (default 8000) |
| output_tokens | number | Average output tokens per request (default 800) |
| cache_hit_ratio | number | Prompt cache hit ratio from 0 to 1 (default 0.75) |
| requests_per_day | number | Daily API requests (default 10000) |
{
"type": "object",
"required": [
"model"
],
"properties": {
"model": {
"type": "string",
"description": "Model ID: gpt-5, claude-4-5-sonnet, gemini-2-5-pro, openai-o3, deepseek-v3-2, or gemini-2-5-flash"
},
"input_tokens": {
"type": "number",
"description": "Average input tokens per request (default 8000)"
},
"output_tokens": {
"type": "number",
"description": "Average output tokens per request (default 800)"
},
"cache_hit_ratio": {
"type": "number",
"description": "Prompt cache hit ratio from 0 to 1 (default 0.75)"
},
"requests_per_day": {
"type": "number",
"description": "Daily API requests (default 10000)"
}
}
}List verified developer, cloud FinOps, and creator calculators with formulas and canonical URLs.
| Argument | Type | Description |
|---|---|---|
| query | string | Optional keyword filter |
{
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Optional keyword filter"
}
}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 11, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.