Gemini 3.1 Flash Lite
Fast & Cheap★ Best ValueGoogle · $0.1/M in · $0.4/M out
$15.6
per month
$0.52/day
AI Pricing Calculator
Estimate your monthly cost across all major AI APIs. Adjust the sliders, see what each model would cost.
Monthly Cost by Model
Sorted cheapest → most expensiveDeepSeek V4
FrontierDeepSeek · $0.27/M in · $1.1/M out
Best price/perf ratio in frontier tier
$42.6
per month
$1.42/day
GPT-5.5 mini
Fast & CheapOpenAI · $0.3/M in · $1.2/M out · $0.03/M cached
$46.8
per month
$1.56/day
Mistral Medium 3.5
BalancedMistral · $0.4/M in · $2/M out
$72
per month
$2.4/day
Llama 4 70B
BalancedMeta (via Together) · $0.88/M in · $0.88/M out
$73.92
per month
$2.46/day
Gemini 3.1 Flash
BalancedGoogle · $0.3/M in · $2.5/M out · $0.08/M cached
$78
per month
$2.6/day
Grok 4 mini
BalancedxAI · $0.5/M in · $2/M out
$78
per month
$2.6/day
Claude Haiku 4.5
Fast & CheapAnthropic · $0.8/M in · $4/M out · $0.08/M cached
$144
per month
$4.8/day
o4-mini
ReasoningOpenAI · $1.1/M in · $4.4/M out
Affordable reasoning model
$171.6
per month
$5.72/day
Mistral Large 3
FrontierMistral · $2/M in · $6/M out
$264
per month
$8.8/day
Gemini 3.1 Pro
FrontierGoogle · $1.25/M in · $10/M out · $0.31/M cached
1M token context
$315
per month
$10.5/day
Llama 4 405B
FrontierMeta (via Together) · $5/M in · $5/M out
Open weights — also runnable locally for free
$420
per month
$14/day
Claude Sonnet 4.6
BalancedAnthropic · $3/M in · $15/M out · $0.3/M cached
$540
per month
$18/day
Grok 4
FrontierxAI · $5/M in · $15/M out
$660
per month
$22/day
GPT-5.5
FrontierOpenAI · $5/M in · $20/M out · $0.5/M cached
$780
per month
$26/day
o3
ReasoningOpenAI · $10/M in · $40/M out
Reasoning model — slower, much more capable on math/logic
$1,560
per month
$52/day
GPT-5.5 Pro
FrontierOpenAI · $15/M in · $60/M out
Higher reasoning, Pro/Business/Enterprise
$2,340
per month
$78/day
Claude Opus 4.7
FrontierAnthropic · $15/M in · $75/M out · $1.5/M cached
$2,700
per month
$90/day
How to use this calculator
Input tokens = everything the model reads (your prompt, attached files, conversation history). Output tokens = what the model writes back.
Rule of thumb: 1 token ≈ 0.75 words. A 1,000-word document is ~1,300 tokens.
Cached input ratio matters because most providers offer steep discounts (10x) on tokens that hit their cache. For applications with consistent system prompts or repeated context, cache hit rates of 50-90% are achievable.
Pricing data is from May 2026. Providers update prices regularly — check official pricing pages before committing to production budgets. Open-weight models like Llama 4 405B and DeepSeek V4 can also be self-hosted for unlimited usage if you have the hardware.