Inference Pricing

BETA

We're still working on our pricing scraper. Please double check prices before making any decisions on this information while we're still in beta.

Example: AI Assistant Chat (10 msg thread)×(users)

Total tokens: 0

Sample Conversation

Input: 0 tokens × 10

Output: 0 tokens × 10

User: I need to build a simple website for my small business. What's the best approach? User: I sell handmade leather goods like wallets, belts, and ...

5 user messages + 5 AI responses

Input Tokens: 0 × 10 = 0

User messages are typically charged at a lower rate

Output Tokens: 0 × 10 = 0

AI responses are typically charged at a higher rate

Jump to Category:

Vendor Filters

Top-tier General-purpose

The most advanced and capable models with strong performance across all tasks. These models excel at complex reasoning, content generation, and understanding nuanced instructions.

Example Use Case:

Building enterprise-grade AI assistants with enhanced reasoning, contextual understanding, and sophisticated task handling capabilities.

Model Name	Provider	Input ($/1M) Price per 1 million tokens (industry standard)	Output ($/1M) Price per 1 million tokens (industry standard)	Example Cost
Claude 3.7 Sonnet	Anthropic	$3.000	$15.000	$0.003
GPT 4.5	OpenAI	$75.000	$150.000	$0.036

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

The most advanced and capable models with strong performance across all tasks. These models excel at complex reasoning, content generation, and understanding nuanced instructions.

Example Use Case:

Building enterprise-grade AI assistants with enhanced reasoning, contextual understanding, and sophisticated task handling capabilities.

Models

Claude 3.7 Sonnet

AnthropicTop-tier General-purpose

Parameters:142B

Input ($/1M):$3.000

Output ($/1M):$15.000

Example Cost:

$0.003

In: $0.000 | Out: $0.003

GPT 4.5

OpenAITop-tier General-purpose

Parameters:200B

Input ($/1M):$75.000

Output ($/1M):$150.000

Example Cost:

$0.036

In: $0.011 | Out: $0.025

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

High-performance

Powerful models offering excellent capabilities with better price-performance ratio than top-tier options. These models handle most complex tasks efficiently.

Example Use Case:

Developing production applications requiring strong reasoning and generation abilities while managing costs, such as automated content creation platforms.

Model Name	Provider	Input ($/1M) Price per 1 million tokens (industry standard)	Output ($/1M) Price per 1 million tokens (industry standard)	Example Cost
Llama 3.1 70B	Inference.net	$0.400	$0.400	$0.000
DeepSeek R1 Distill Llama 70B (FP8)	Inference.net	$0.400	$0.400	$0.000
Llama 3.3 70B	Inference.net	$0.400	$0.400	$0.000
Llama 3 70B	Groq	$0.590	$0.790	$0.000
Qwen2-72B	Together AI	$0.900	$0.900	$0.000
Llama 3 70B	Together AI	$0.900	$0.900	$0.000
DeepSeek V3 (FP8)	Inference.net	$1.200	$1.200	$0.000
GPT-4o	OpenAI	$2.500	$10.000	$0.002

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Powerful models offering excellent capabilities with better price-performance ratio than top-tier options. These models handle most complex tasks efficiently.

Example Use Case:

Developing production applications requiring strong reasoning and generation abilities while managing costs, such as automated content creation platforms.

Models

Llama 3.1 70B

Inference.netHigh-performance

Parameters:70B

Input ($/1M):$0.400

Output ($/1M):$0.400

Example Cost:

$0.000

In: $0.000 | Out: $0.000

DeepSeek R1 Distill Llama 70B (FP8)

Inference.netHigh-performance

Parameters:70B

Input ($/1M):$0.400

Output ($/1M):$0.400

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3.3 70B

Inference.netHigh-performance

Parameters:70B

Input ($/1M):$0.400

Output ($/1M):$0.400

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3 70B

GroqHigh-performance

Parameters:70B

Input ($/1M):$0.590

Output ($/1M):$0.790

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Qwen2-72B

Together AIHigh-performance

Parameters:72B

Input ($/1M):$0.900

Output ($/1M):$0.900

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3 70B

Together AIHigh-performance

Parameters:70B

Input ($/1M):$0.900

Output ($/1M):$0.900

Example Cost:

$0.000

In: $0.000 | Out: $0.000

DeepSeek V3 (FP8)

Inference.netHigh-performance

Parameters:127B

Input ($/1M):$1.200

Output ($/1M):$1.200

Example Cost:

$0.000

In: $0.000 | Out: $0.000

GPT-4o

OpenAIHigh-performance

Parameters:1.5T

Input ($/1M):$2.500

Output ($/1M):$10.000

Example Cost:

$0.002

In: $0.000 | Out: $0.002

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Mid-range

Balanced models offering good capabilities at moderate costs. These models handle common tasks well and are suitable for most general applications.

Example Use Case:

Creating customer support chatbots and knowledge retrieval systems where cost efficiency and good performance are both important.

Model Name	Provider	Input ($/1M) Price per 1 million tokens (industry standard)	Output ($/1M) Price per 1 million tokens (industry standard)	Example Cost
Llama 3.2 11B Vision Instruct (FP16)	Inference.net	$0.055	$0.055	$0.000
Mistral Nemo 12B Instruct (FP8)	Inference.net	$0.100	$0.100	$0.000
ChatGPT 3o-mini	OpenAI	$0.150	$0.600	$0.000
Gemini 2.0 Flash	Google	$0.150	$0.600	$0.000
Claude 3.5 Haiku	Anthropic	$0.800	$4.000	$0.001

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Balanced models offering good capabilities at moderate costs. These models handle common tasks well and are suitable for most general applications.

Example Use Case:

Creating customer support chatbots and knowledge retrieval systems where cost efficiency and good performance are both important.

Models

Llama 3.2 11B Vision Instruct (FP16)

Inference.netMid-range

Parameters:11B

Input ($/1M):$0.055

Output ($/1M):$0.055

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Mistral Nemo 12B Instruct (FP8)

Inference.netMid-range

Parameters:12B

Input ($/1M):$0.100

Output ($/1M):$0.100

Example Cost:

$0.000

In: $0.000 | Out: $0.000

ChatGPT 3o-mini

OpenAIMid-range

Parameters:25B

Input ($/1M):$0.150

Output ($/1M):$0.600

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Gemini 2.0 Flash

GoogleMid-range

Parameters:100B

Input ($/1M):$0.150

Output ($/1M):$0.600

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Claude 3.5 Haiku

AnthropicMid-range

Parameters:20B

Input ($/1M):$0.800

Output ($/1M):$4.000

Example Cost:

$0.001

In: $0.000 | Out: $0.001

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Cost-effective

Budget-friendly models optimized for efficiency and lower costs. These models perform well on straightforward tasks while minimizing token usage expenses.

Example Use Case:

Building high-volume applications like content tagging, classification systems, or user intent recognition where scale matters.

Model Name	Provider	Input ($/1M) Price per 1 million tokens (industry standard)	Output ($/1M) Price per 1 million tokens (industry standard)	Example Cost
Llama 3.2 1B Instruct (FP16)	Inference.net	$0.010	$0.010	$0.000
Llama 3.2 3B Instruct	Inference.net	$0.020	$0.020	$0.000
Llama 3.1 8B Instruct (FP8)	Inference.net	$0.025	$0.025	$0.000
Llama 3.1 8B Instruct (FP16)	Inference.net	$0.030	$0.030	$0.000
GPT-4o Mini	OpenAI	$0.150	$0.600	$0.000
Llama 3.1 7B	Together AI	$0.200	$0.200	$0.000

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Budget-friendly models optimized for efficiency and lower costs. These models perform well on straightforward tasks while minimizing token usage expenses.

Example Use Case:

Building high-volume applications like content tagging, classification systems, or user intent recognition where scale matters.

Models

Llama 3.2 1B Instruct (FP16)

Inference.netCost-effective

Parameters:1B

Input ($/1M):$0.010

Output ($/1M):$0.010

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3.2 3B Instruct

Inference.netCost-effective

Parameters:3B

Input ($/1M):$0.020

Output ($/1M):$0.020

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3.1 8B Instruct (FP8)

Inference.netCost-effective

Parameters:8B

Input ($/1M):$0.025

Output ($/1M):$0.025

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3.1 8B Instruct (FP16)

Inference.netCost-effective

Parameters:8B

Input ($/1M):$0.030

Output ($/1M):$0.030

Example Cost:

$0.000

In: $0.000 | Out: $0.000

GPT-4o Mini

OpenAICost-effective

Parameters:250B

Input ($/1M):$0.150

Output ($/1M):$0.600

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Llama 3.1 7B

Together AICost-effective

Parameters:7B

Input ($/1M):$0.200

Output ($/1M):$0.200

Example Cost:

$0.000

In: $0.000 | Out: $0.000

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Reasoning-focused

Models specialized in logical reasoning, problem-solving, and analytical tasks. These models excel at tasks requiring step-by-step thinking and mathematical operations.

Example Use Case:

Implementing code generation tools, math problem solvers, or data analysis assistants requiring precise logical reasoning.

Model Name	Provider	Input ($/1M) Price per 1 million tokens (industry standard)	Output ($/1M) Price per 1 million tokens (industry standard)	Example Cost
Qwen 2.5 72B Instruct (FP8)	Inference.net	$0.350	$0.350	$0.000
DeepSeek R1 (FP8)	Inference.net	$3.000	$3.000	$0.001
DeepSeek R1	Together AI	$3.000	$7.000	$0.002
Claude 3.7 Sonnet Thinking	Anthropic	$8.000	$24.000	$0.005

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Models specialized in logical reasoning, problem-solving, and analytical tasks. These models excel at tasks requiring step-by-step thinking and mathematical operations.

Example Use Case:

Implementing code generation tools, math problem solvers, or data analysis assistants requiring precise logical reasoning.

Models

Qwen 2.5 72B Instruct (FP8)

Inference.netReasoning-focused

Parameters:72B

Input ($/1M):$0.350

Output ($/1M):$0.350

Example Cost:

$0.000

In: $0.000 | Out: $0.000

DeepSeek R1 (FP8)

Inference.netReasoning-focused

Parameters:236B

Input ($/1M):$3.000

Output ($/1M):$3.000

Example Cost:

$0.001

In: $0.000 | Out: $0.001

DeepSeek R1

Together AIReasoning-focused

Parameters:172B

Input ($/1M):$3.000

Output ($/1M):$7.000

Example Cost:

$0.002

In: $0.000 | Out: $0.001

Claude 3.7 Sonnet Thinking

AnthropicReasoning-focused

Parameters:142B

Input ($/1M):$8.000

Output ($/1M):$24.000

Example Cost:

$0.005

In: $0.001 | Out: $0.004

Pricing shown per 1M tokens. Example costs are estimates only and may vary based on actual tokenization.

Last updated: 7/8/2025

Download all available models and pricing data instantly.

Navigate the complex landscape of AI models with ease. This tool helps developers find the right model for their projects by comparing pricing across providers. We are also working on improving our model categorizations. If you have suggestions for how we can improve our grouping or tags/categories, please create an issue in our GitHub repo.