What is the cheapest AI API in 2026?

Qwen 3.7 Flash is the cheapest at $0.03/$0.13 per million input/output tokens with vision support. GPT-5 nano ($0.05/$0.40) and GPT-oss 20B ($0.08/$0.35) are the cheapest OpenAI options. DeepSeek V4 Flash ($0.22/$0.66 off-peak) and Gemini 2.5 Flash-Lite ($0.10/$0.40) are also strong budget picks.

How much does GPT-5 cost per API call?

GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens. A typical 1000-token request costs about $0.00125 for input and $0.01 for output. GPT-5 mini is 5x cheaper at $0.25/$2.00.

How do I calculate my monthly AI API costs?

Multiply your daily requests by average tokens per request, then by the model's per-token price. Use the APIpulse calculator above to estimate costs across 95 models instantly — just enter your request volume and token usage.

Which AI provider is cheapest for API access?

Alibaba/Qwen, Google, and OpenAI offer the cheapest models. Qwen 3.7 Flash ($0.03/$0.13) is the absolute cheapest with vision. GPT-5 nano ($0.05/$0.40) is cheapest from OpenAI. DeepSeek V4 Flash ($0.22/$0.66 off-peak) and Gemini 2.5 Flash-Lite ($0.10/$0.40) are strong budget options. For mid-tier, Mistral Large 3 ($0.50/$1.50) offers excellent value.

AI API Cost Calculator

Compare pricing across 95 models from 11 providers. Enter your usage below to see exactly what you'd pay — no signup required.

Typical request:

By volume:

API mode:

Provider & Model

Input tokens per request

Output tokens per request

Requests per day

Days per month

Cost Estimate

Cost per request $0.0000

Cost per 1K requests $0.00

Input cost (monthly) $0.00

Output cost (monthly) $0.00

Total tokens/month 0

Monthly Total $0.00

Based on published API pricing. Actual costs may vary.

See how prices have changed →

1-5 request types · 6-8 volume · C copy · ? shortcuts

How This Calculator Works

Enter your expected usage — tokens per request, requests per day, and days per month — and the calculator shows your estimated cost per request, cost per 1K requests, and monthly total for any of 95 AI models across 11 providers. Use the "Typical request" presets to quickly see what common workloads cost. It also suggests the cheapest alternative if a cheaper model can handle your workload.

Supported Providers

OpenAI (27 models): GPT-5.6 Sol/Terra/Luna, GPT-5.5, GPT-5.4, GPT-5.3 Codex, GPT-5.2/5.1/5, GPT-5 Pro/nano, GPT-4.1 family, o3/o3 Pro/o4-mini, GPT-oss 120B/20B
Anthropic (10 models): Opus 5, Fable 5, Sonnet 5, Haiku 4.5, Opus 4.8/4.7, Sonnet 4.6/4.5
Google (14 models): Gemini 3.1 Pro, Gemini 3.7/3.6/3.5 Flash, Gemini 3 Flash, Gemini 3.5/3.1/2.5 Flash-Lite, Gemini 2.5 Pro/Flash, Nano Banana 2/Pro
DeepSeek (3 models): V4 Pro, V4 Flash, V3.2
Mistral (11 models): Large 3, Medium 3.5, Small 4, Codestral, Ministral 3 3B/8B/14B, Magistral Medium/Small (retired)
Alibaba/Qwen (1 model): Qwen 3.7 Flash (cheapest multimodal)
Cohere (3 models): Command R+, Command R, Command A
Meta (Together.ai) (3 models): Llama 4 Scout, Llama 4 Maverick, Llama 3.3 70B
Moonshot (3 models): Kimi K3, Kimi K2.6, Kimi K2.7 Code
xAI (5 models): Grok 4.6, Grok 4.5, Grok 4.20, Grok 4.3, Grok Build 0.1
AI21 (2 models): Jamba 1.7 Large, Jamba Mini

Common Use Cases

Our calculator helps you estimate costs for:

Chatbots and virtual assistants — typically 500-2000 input tokens, 200-800 output tokens per turn
Code generation — 1000-5000 input tokens, 500-3000 output tokens per request
Document analysis — 3000-10000 input tokens, 200-1000 output tokens per document
Content generation — 500-2000 input tokens, 1000-4000 output tokens per article

Want to compare two models side by side?

Use the Comparison Tool

Tips to Reduce Your API Costs

The calculator often reveals that a cheaper model can handle your workload. Here are the biggest cost-saving strategies:

Use budget models for simple tasks: GPT-5.4 nano ($0.20/$1.25) handles 80% of chatbot requests at 1/25th the cost of GPT-5.5
Optimize prompts: Shorter prompts = fewer input tokens = lower costs
Set token limits: Don't let models generate 4000 tokens when 200 will do
Cache responses: Identical prompts can be cached to eliminate redundant API calls

Read our full guide: How to Cut Your AI API Bill in Half: 10 Practical Tips

Get weekly AI pricing updates & cost-saving tips

Join 200+ developers optimizing their AI API spend. No spam, unsubscribe anytime.

Provider Calculators

Cohere Cost Calculator — Command R+ & Command R pricing
Moonshot Cost Calculator — Kimi K3 & K2.6 pricing
Together.ai Cost Calculator — Llama 4 & Llama 3.1 pricing