Skip to content
Go back

Coding-Agent Pricing Simulator

Coding-agent pricing simulator

Shape a conversation and compare cached and uncached LLM call costs.

Pricing

Dollars per million tokens

Model preset
$
$
$
$

Workflow

Conversation shape and tool activity

3
12
Tool mix
50%
30%
20%
Enable cache

Results

Total cost

$1.19

$1.19 with cache · $6.44 without · cache saves 82%

Cost split

Cached input 50%Cache write 28%Uncached input 0%Output 23%
User turns
3
LLM calls
39
Uncached input tokens
0
Cache write tokens
52.7k
Cache read tokens
1.2M
Output tokens
10.8k

Cost per call

Each stacked bar is one LLM call. Its colors show what that call spent on cached input, cache writes, uncached input, and output. The first bar contains the system prompt (which is usually rather large).

Per-call operation costs$0.00$0.04$0.07Cost per callLLM call 1Call 39
Cached input: $0.59Cache write: $0.33Uncached input: $0.00Output: $0.27
User query

Cumulative cost

The colored bands add each operation over the conversation. The dark line is the running total.

Cumulative operation and total costs$0.00$0.30$0.60$0.89$1.19Cumulative costLLM call 1Call 39
Cached input: $0.59Cache write: $0.33Uncached input: $0.00Output: $0.27Total: $1.19

Provider-specific TTLs, routing misses, cache breakpoints, and block lookback are not modeled.


Share this post on:

Next Post
From Next-Word Predictor to Copilots