AI Model Cost Comparison Database (2026)

API pricing for 16 major LLM models. Input/output per million tokens, context windows, latency, and daily cost at 10K requests. CC BY 4.0.

Download CSV (16 models)
ModelProviderInputOutputContextSpeedDaily Cost (10K req)
GPT-4oOpenAI$2.50$10.00128KMedium$75
GPT-4o-miniOpenAI$0.15$0.60128KFast$4.50
Claude 3.5 SonnetAnthropic$3.00$15.00200KMedium$90
Claude 3 HaikuAnthropic$0.25$1.25200KFast$7.50
Gemini 1.5 ProGoogle$1.25$5.002MMedium$37.50
Gemini 1.5 FlashGoogle$0.075$0.301MFast$2.25
GPT-4 TurboOpenAI$10.00$30.00128KSlow$225
o1OpenAI$15.00$60.00200KSlow$450
Llama 3.1 70BMeta/Groq$0.59$0.79128KVery Fast$14.40
Llama 3.1 405BMeta$3.00$3.00128KMedium$30
DeepSeek V3DeepSeek$0.27$1.10128KFast$8.20
Mistral LargeMistral$2.00$6.00128KMedium$45
Command R+Cohere$2.50$10.00128KMedium$75
Nova ProAmazon$0.80$3.20300KMedium$24
Nova LiteAmazon$0.06$0.24300KFast$1.80
Cite this data:
sipi.bot. "AI Model Cost Comparison Database," 2026. CC BY 4.0.
CSV: ai-model-costs-2026.csv

Why This Matters for AI Agents

An autonomous AI agent making 10,000 API calls per day on GPT-4o costs $75/day ($2,250/month). The same workload on Gemini 1.5 Flash costs $2.25/day ($67/month). A spend firewall like sipi.bot monitors these costs in real time, blocking runaway agents before they drain your budget. The 33x price difference between the cheapest and most expensive models is why cost control is the #1 operational concern for agent builders in 2026.