What is the cheapest AI API in 2026?

The cheapest AI APIs in 2026 are Llama 3.1 8B ($0.10/$0.10 per 1M tokens), GPT-oss 20B ($0.08/$0.35), and Mistral Small 4 ($0.15/$0.60). DeepSeek V4 Flash ($0.14/$0.28) offers the best value for a 1M context window.

How much does GPT-5 cost per API call?

GPT-5 costs $1.25 per 1M input tokens and $10.00 per 1M output tokens. A typical 1,000-token chat interaction costs about $0.011. GPT-5 mini is much cheaper at $0.25/$2.00 per 1M tokens.

Which AI API gives the most value for money?

For budget-conscious developers, DeepSeek V4 Pro ($0.43/$0.87 per 1M tokens with 1M context) and GPT-5 mini ($0.25/$2.00) offer the best value. For premium quality, Claude Sonnet 4.6 ($3.00/$15.00) and GPT-5 ($1.25/$10.00) deliver strong performance per dollar.

📊 AI API Pricing Cheat Sheet 2026

Q: What are hidden costs of AI APIs?

Three hidden costs catch developers off guard: (1) Output token markup — outputs cost 3-20x more than inputs across all providers. (2) Context window waste — sending full conversation history on every call multiplies input costs. (3) Retry and error costs — failed API calls still consume tokens for partial processing.

Every model. Every provider. One page. Updated weekly.

Last updated: July 7, 2026 · 88 models · 10 providers

✓ 38 Active Models 10 Providers 37× Price Range

$0.07

Cheapest Input (per 1M tok)

$0.10

Cheapest Output (per 1M tok)

$180

Most Expensive Output

1.05M

Largest Context Window

📋 Full Pricing Table — All 59 Models

Click any column header to sort. Prices are per 1 million tokens. Filter by provider or tier.

Model	Provider	Tier	Input $/1M	Output $/1M	Blended*	Context

* Blended = 3:1 output:input ratio (typical for chat). Click column headers to sort. Greyed rows are deprecated.

🏢 Provider Comparison

At a glance: what each provider offers and where they shine.

OpenAI

9 models (7 active)

Range: $0.08 — $180/1M output

Strength: Broadest lineup

GPT-oss 20B → GPT-5.5 Pro

Anthropic

6 models (4 active)

Range: $5 — $75/1M output

Strength: Code & reasoning

Haiku 4.5 → Claude 4 Opus

Google

8 models (6 active)

Range: $0.30 — $12/1M output

Strength: Best value mid-tier

Flash Lite → Gemini 3.1 Pro

DeepSeek

4 models (3 active)

Range: $0.28 — $0.87/1M output

Strength: Cheapest quality

V4 Flash → V4 Pro

Mistral

3 models (3 active)

Range: $0.30 — $7.50/1M output

Strength: EU compliance

Small 4 → Large 3

Meta (via Together)

4 models (3 active)

Range: $0.10 — $0.88/1M output

Strength: Open-source pricing

Llama 3.1 8B → Llama 4

Cohere

3 models (3 active)

Range: $1.50 — $10/1M output

Strength: Enterprise RAG

Command R → Command A

xAI

2 models (2 active)

Range: $0.50 — $2.50/1M output

Strength: Large context

Grok Build → Grok 4.3

Moonshot

1 model

Range: $4.00/1M output

Strength: Chinese market

Kimi K2.6

AI21

2 models (1 active)

Range: $8.00/1M output

Strength: Long-context SSM

Jamba 1.7 Large

💰 Cost Per Task — What You'll Actually Pay

Real-world cost estimates for common AI tasks. Based on typical token counts.

💬 Simple Chat

~500 input + ~300 output tokens

Llama 3.1 8B$0.0001

GPT-5 mini$0.001

Claude Sonnet 4.6$0.006

GPT-5$0.004

Claude Opus 4.8$0.010

📄 Summarize Document

~4,000 input + ~500 output tokens

Llama 3.1 8B$0.0005

GPT-5 mini$0.002

Claude Sonnet 4.6$0.020

GPT-5$0.010

Claude Opus 4.8$0.033

🔧 Generate Code

~1,000 input + ~2,000 output tokens

Llama 3.1 8B$0.0003

GPT-5 mini$0.004

Claude Sonnet 4.6$0.033

GPT-5$0.021

Claude Opus 4.8$0.055

📊 Analyze Data

~10,000 input + ~1,000 output tokens

Llama 3.1 8B$0.001

GPT-5 mini$0.005

Claude Sonnet 4.6$0.045

GPT-5$0.023

Claude Opus 4.8$0.075

🤖 AI Agent Step

~2,000 input + ~1,500 output tokens

Llama 3.1 8B$0.0004

GPT-5 mini$0.004

Claude Sonnet 4.6$0.029

GPT-5$0.018

Claude Opus 4.8$0.048

📚 RAG Query

~15,000 input + ~500 output tokens

Llama 3.1 8B$0.002

GPT-5 mini$0.005

Claude Sonnet 4.6$0.053

GPT-5$0.024

Claude Opus 4.8$0.088

💡 Key insight: Output tokens cost 3–20× more than input tokens across all providers. Design your prompts to minimize output length where possible.

🎯 Quick Recommendations by Use Case

Not sure which model to pick? Start here.

💸 Lowest Cost

When budget is everything

Llama 3.1 8B

$0.10 / $0.10 per 1M

⚖️ Best Value

Quality + affordability

DeepSeek V4 Pro

$0.43 / $0.87 per 1M

🧠 Best Reasoning

Complex analysis & coding

Claude Opus 4.8

$5.00 / $25.00 per 1M

⚡ Best Speed

Real-time apps, chatbots

Gemini 3.5 Flash

$1.50 / $9.00 per 1M

📝 Long Documents

Large context window needed

GPT-5.5

$5.00 / $30.00 per 1M (1.05M ctx)

💻 Code Generation

Coding copilots & IDE tools

Claude Sonnet 4.6

$3.00 / $15.00 per 1M

🏢 Enterprise RAG

Retrieval-augmented generation

Command R+

$2.50 / $10.00 per 1M

🌍 EU Compliance

GDPR-friendly, EU hosting

Mistral Large 3

$0.50 / $1.50 per 1M

🚀 Budget General

Good enough for most tasks

GPT-5 mini

$0.25 / $2.00 per 1M

⚠️ 3 Hidden Costs Most Developers Miss

The per-token price is just the beginning. Here's what catches teams off guard.

1. Output Token Markup (3–20×)

Every provider charges more for output tokens than input. The markup ranges from 3× (DeepSeek) to 20× (GPT-5.5 Pro). For code generation and long-form writing, output tokens dominate your bill.

💡 Tip: Set max_tokens limits. Use structured output (JSON mode) to reduce verbose responses.

2. Context Window Waste

Sending full conversation history on every API call means your input costs grow linearly with conversation length. A 50-turn chat with GPT-5 at 10K context costs $0.125 per message just for input.

💡 Tip: Implement conversation summarization. Trim old messages. Use sliding window approaches.

3. Retry & Error Costs

Rate limits, timeouts, and malformed responses all consume tokens. With aggressive retry logic, you can accidentally 3× your costs. Some providers charge for partial completions.

💡 Tip: Implement exponential backoff. Set retry budgets. Monitor error rates per provider.

Want detailed cost reports for your exact usage?

APIpulse includes monthly cost tracking, provider comparison reports, and optimization recommendations tailored to your API usage patterns.

Explore Free Tools →

100% free · No signup · No credit card

📊 AI API Pricing Cheat Sheet 2026

📋 Full Pricing Table — All 59 Models

🏢 Provider Comparison

OpenAI

Anthropic

Google

DeepSeek

Mistral

Meta (via Together)

Cohere

xAI

Moonshot

AI21

💰 Cost Per Task — What You'll Actually Pay

💬 Simple Chat

📄 Summarize Document

🔧 Generate Code

📊 Analyze Data

🤖 AI Agent Step

📚 RAG Query

🎯 Quick Recommendations by Use Case

💸 Lowest Cost

⚖️ Best Value

🧠 Best Reasoning

⚡ Best Speed

📝 Long Documents

💻 Code Generation

🏢 Enterprise RAG

🌍 EU Compliance

🚀 Budget General

⚠️ 3 Hidden Costs Most Developers Miss

1. Output Token Markup (3–20×)

2. Context Window Waste

3. Retry & Error Costs

Want detailed cost reports for your exact usage?

Related Tools

🧮 Cost Calculator

⚖️ Compare Models

🎯 AI API Advisor

📊 2026 Benchmark

🔀 Model Switch

🔍 Model Finder

📡 Pricing API

👥 Team Tools