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
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).
Cached input: $0.59Cache write: $0.33Uncached input: $0.00Output: $0.27
User queryCumulative cost
The colored bands add each operation over the conversation. The dark line is the running total.
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.