Reasoning Models
Estimate reasoning model API costs for agents, coding, and long-context tasks across input, output, cache-read, and monthly request volume.
OpenAI · GPT-5.4 mini
$0.00
per request
Input $0.75/1M tokens
Output $4.5/1M tokens
Cache read $0.075/1M tokens
Input0%
Output0%
Cache read0%
Anthropic · Claude Sonnet 4.6
$0.00
per request
Input $3/1M tokens
Output $15/1M tokens
Cache write $3.75/1M tokens
Cache read $0.3/1M tokens
Uncached0%
Cache write0%
Cache read0%
Output0%
DeepSeek · DeepSeek V4 Pro
$0.00
per request
Cache miss $0.14/1M tokens
Output $0.28/1M tokens
Cache hit $0.0028/1M tokens
Cache miss0%
Output0%
Cache hit0%
Reasoning API cost decisions and model routing
Reasoning model ROI vs text model substitution
Compare task complexity, failure cost, and output quality to decide when reasoning models are worth the premium.
AI Agent multi-step call cost modeling
Break tool calls, loops, retries, and intermediate context into budgetable Agent cost fields.
Low-cost model routing and escalation strategy
Design routing that starts cheap, escalates failures, and reserves stronger models for high-value tasks.