Posted by Introduction to GPT-4o Update
On May 13, 2024, OpenAI announced its new flagship model, GPT-4o. GPT-4o is an innovative multimodal model capable of processing text, audio, and image data in real-time, making it faster and more cost-efficient than previous models. This update has generated significant anticipation among ChatGPT users, particularly API users, as the changes in pricing policies are expected to have a substantial impact.
The Need for New Pricing Policies
With the introduction of GPT-4o, the performance and efficiency of ChatGPT have greatly improved. Consequently, OpenAI has introduced new pricing policies to ensure that more users can access high-performance AI at a reasonable cost. Additionally, the new tokenization technology applied in the GPT-4o update enhances text processing efficiency, leading to cost savings. This pricing policy change reflects OpenAI’s goal of providing better value to both users and developers.
Key Changes with GPT-4o Update
Reduced API Usage Costs
GPT-4o’s API usage costs are 50% lower than those of GPT-4 Turbo. While GPT-4 Turbo maintains its existing pricing ($10.0 /1M tokens), GPT-4o is priced more affordably ($5.0 /1M tokens). This aims to enable more developers to utilize high-performance AI without financial burdens and to integrate ChatGPT into a variety of applications. For example, compared to using GPT-4 Turbo, developers can handle twice as many requests within the same budget when using GPT-4o.
Improved Processing Speed
GPT-4o boasts faster processing speeds than its predecessors, meaning more tasks can be completed in a shorter time, significantly enhancing user experience. Faster response times are particularly crucial for real-time applications such as chatbots, real-time data analysis, and speech recognition applications, providing a significant advantage.
Support for Multimodal Capabilities
The new pricing policy includes not only text but also audio and image data processing. This allows users to handle various forms of data integratively, experiencing richer and more intuitive interactions. This is especially useful for multimedia content creation, complex data analysis, and visual data processing applications.
Changes in Free and Paid Plans
| Plan | Price | Key Features | Message Limit | Multimodal Capabilities Usage |
|---|---|---|---|---|
| Free Plan | Free | Text and image processing | Basic limit provided | Limited use |
| Plus Plan | $20/month | All features available | Up to 5x message limit | Priority use |
| Team Plan | $30/month (per user) | Team management tools, integration | Increased allocation per user | Multimodal capabilities for team collaboration |
| Enterprise Plan | Custom pricing | Advanced features, custom solutions | Custom allocation | Enterprise-level multimodal capabilities |
Free Plan($0.0 /month)
Free users primarily use GPT-3.5 and can now access GPT-4o with certain limitations. This measure aims to allow more users to experience the latest AI technology. Free plan users can process a limited number of requests each month for free and experience the basic features of high-performance AI at no cost.
Plus Plan($20.0 /month)
Plus plan users have access to all GPT-4o features, with up to 5x more message limits. They can also use previous versions like GPT-4 and GPT-3.5. The Plus Plan offers significant benefits for frequent ChatGPT users, providing priority access to multimodal capabilities like audio data processing, for a monthly fee of $20.0.
Team Plan($30.0 /month per user, $25.0 /month billed annually)
The Team Plan is designed for small teams or enterprises, allowing team members to collaborate using GPT-4o’s features. It includes the benefits of the Plus Plan, along with increased allocations and additional team management tools and integration features. This plan facilitates team collaboration and project management, enhancing productivity. The monthly fee is $30.0 per user, with discounted rates for annual billing.
Enterprise Plan
The Enterprise Plan targets large enterprises, offering advanced features and customized solutions. It includes the benefits of the Team Plan, supports large-scale data processing and complex workflows, and provides enterprise-grade security and support services. The Enterprise Plan is customizable to meet specific client needs, including a dedicated account manager and 24/7 support. Pricing is tailored based on the company’s size and requirements.
API Pricing Comparison
Language Model Pricing
GPT-4o
- gpt-4o Models: Input $5.0 / 1M tokens, Output $15.0 / 1M tokens (50% cheaper than GPT-4 Turbo)
(gpt-4o, gpt-4o-2024-05-13 models) - Vision pricing (1024 px x 1024 px): $0.003825
GPT-4 Turbo
- gpt-4-turbo Models: Input $10.0 / 1M tokens, Output $30.0 / 1M tokens
(gpt-4-turbo, gpt-4-turbo-2024-04-09 models) - Vision pricing (1024 px x 1024 px): $0.00765
GPT-4
- gpt-4 Model: Input $30.0 / 1M tokens, Output $60.0 / 1M tokens
- gpt-4-32k Model: Input $60.0 / 1M tokens, Output $120.0 / 1M tokens
GPT-3.5 Turbo
- gpt-3.5-turbo-0125 Model: Input $0.5 / 1M tokens, Output $1.5 / 1M tokens
- gpt-3.5-turbo-instruct Model: Input $1.5 / 1M tokens, Output $2.0 / 1M tokens
Assistants API
- Code interpreter: $0.03 / session
- File Search: $0.10 / GB of vector-storage per day (1GB free)
Fine-tuning Models
- gpt-3.5-turbo Model: Training $8.0 / 1M tokens, Input $3.0 / 1M tokens, Output $6.0 / 1M tokens
- davinci-002 Model: Training $8.0 / 1M tokens, Input $3.0 / 1M tokens, Output $6.0 / 1M tokens
- babbage-002 Model: Training $0.4 / 1M tokens, Input $1.6 / 1M tokens, Output $1.6 / 1M tokens
Embedding Models
- text-embedding-3-small Model: $0.02 / 1M tokens
- text-embedding-3-large Model: $0.13 / 1M tokens
- ada v2 Model: $0.10 / 1M tokens
Base Models
- davinci-002 Model: $2.0 / 1M tokens
- babbage-002 Model: $0.4 / 1M tokens
Image Model Pricing
Image Models
- DALL-E3 Standard quality – 1024 x 1024: $0.04 / image, 1024 x 1792 (1792 x 1024): $0.08 / image
- DALL-E3 HD quality – 1024 x 1024: $0.08 / image, 1024 x 1792 (1792 x 1024): $0.12 / image
- DALL-E2 – 1024 x 1024: $0.02 / image, 512 x 512: $0.018 / image, 256 x 256: $0.016 / image
Audio Model Pricing
Audio Models
- Whisper: $0.006 / minute
- TTS: $15.0 / 1M characters
- TTS HD: $30.0 / 1M characters
Token and Cost Reduction Strategies
The introduction of OpenAI’s latest model, GPT-4o, focuses on enhancing AI accessibility and providing cost-efficient solutions for both users and developers. This section delves deeper into token and cost reduction strategies.
Token Optimization
GPT-4o introduces new tokenization technology, significantly reducing the number of tokens across various languages. This improves data compression efficiency, making text processing more efficient and ultimately leading to cost savings.
- Gujarati: Reduced from 145 to 33 tokens (4.4x reduction)
- Telugu: Reduced from 159 to 45 tokens (3.5x reduction)
- Tamil: Reduced from 116 to 35 tokens (3.3x reduction)
- Marathi: Reduced from 96 to 33 tokens (2.9x reduction)
- Hindi: Reduced from 90 to 31 tokens (2.9x reduction)
- Urdu: Reduced from 82 to 33 tokens (2.5x reduction)
- Arabic: Reduced from 53 to 26 tokens (2.0x reduction)
- Persian: Reduced from 61 to 32 tokens (1.9x reduction)
- Russian: Reduced from 39 to 23 tokens (1.7x reduction)
- Korean: Reduced from 45 to 27 tokens (1.7x reduction)
- Vietnamese: Reduced from 46 to 30 tokens (1.5x reduction)
- Chinese: Reduced from 34 to 24 tokens (1.4x reduction)
- Japanese: Reduced from 37 to 26 tokens (1.4x reduction)
- Turkish: Reduced from 39 to 30 tokens (1.3x reduction)
- Italian: Reduced from 34 to 28 tokens (1.2x reduction)
- German: Reduced from 34 to 29 tokens (1.2x reduction)
- Spanish: Reduced from 29 to 26 tokens (1.1x reduction)
- Portuguese: Reduced from 30 to 27 tokens (1.1x reduction)
- French: Reduced from 31 to 28 tokens (1.1x reduction)
- English: Reduced from 27 to 24 tokens (1.1x reduction)
This reduction in token count enhances the efficiency of text data and contributes significantly to cost savings.
Cost Reduction Strategies
GPT-4o reduces API usage costs by half compared to previous models, providing substantial cost savings for users and developers. Here are additional cost-saving strategies for utilizing GPT-4o:
Efficient API Usage:
- Minimize Request Count: Reduce the number of requests needed by optimizing questions to obtain more information in a single request.
- Cache Results: Cache results for repeated requests to reduce redundant API calls.
Token Usage Optimization:
- Concise Inputs: Use concise inputs with only essential information to reduce the number of tokens.
Cost-Efficient Model Selection:
- Appropriate Model Usage: Choose the appropriate model based on task complexity. For instance, use GPT-3.5 Turbo for simple tasks and GPT-4o for complex tasks.
- Translation Strategy: To save costs, translate complex questions and answers into more concise languages. For example, Korean text uses more tokens compared to English. By translating content to English for processing, you can reduce token usage and save costs.
You can check the token count using the OpenAI – Tokenizer(the GPT-4o Tokenizer has not been released yet).
By implementing these strategies, you can effectively manage AI utilization costs.
Expected Benefits
The new pricing policies, especially those for GPT-4o, are expected to provide better value to users and increase AI technology accessibility. Developers can always choose and utilize cost-efficient, high-performance AI to develop various innovative applications. Additionally, the fast processing speed and multimodal capabilities will greatly enhance user experience.