Support for Low-Cost API Models in Too Code VSCode Extension - GPTAPI.US

[benzntech]

Description

I propose adding API support for various low-cost AI models such as ChatGPT, Claude, DeepSeek, Gemini, and more to the Too Code VSCode extension. This will enable users to access cost-effective AI services for code generation, assistance, and real-time interactions.

---

Key Features Requested

1. API Integration for Various AI Models

   • Support AI providers such as OpenAI (GPT-4o), Claude (Anthropic), DeepSeek, Gemini, and others.
   • Enable API-based communication with these models through a simple configuration file.

2. Online Recharge & Chat

   • Users should be able to input their API keys directly in the extension settings.
   • Provide a status monitoring system to track token usage and request failures.

3. Shortened Model Naming Support

   • Allow shorthand naming for models, similar to how Claude-3-5-Sonnet can be used as c-3-5-sonnet in some platforms.
   • Standardized model naming for easy switching between providers.

4. Model-Specific Integration Requirements

   • ChatGPT (OpenAI) Support:
     • Full support for file transfer, networking, DALL·E 3 image generation, and voice intercom.
   • Claude Models:
     • Ensure Claude Opus and Sonnet versions support image transfer (Base64/URL).
   • Gemini API Usage:
     • Follow OpenAI's model structure (e.g., api-gemini-2.0-flash-exp in NexChat).
   • DeepSeek & Other Models:
     • Ensure pricing and concurrency support match the provider specifications.

5. Custom API Configuration

   • Users should be able to specify their preferred API endpoints.
   • For example:
     • OpenAI mode: Replace api.openai.com with api.gptapi.us.
     • Claude mode: Replace api.anthropic.com with api.gptapi.us.

---

Pricing & Cost Consideration

Given the fluctuating nature of AI model pricing, a built-in cost estimator could be beneficial. Users could view token consumption rates before executing large queries.

• Example pricing (converted for clarity):
  • GPT-4o: $0.0025 / 1K tokens (input), $0.01 / 1K tokens (output)
  • Claude 3.5 Sonnet: $0.003 / 1K tokens (input), $0.015 / 1K tokens (output)
  • DeepSeek-v3: $0.00014 / 1K tokens (input), $0.00028 / 1K tokens (output)
  • Discounted rates: Some services offer up to 75% off on bulk usage.

---

Potential Benefits of This Feature

• Cost Efficiency: Users can choose lower-cost AI services instead of being restricted to expensive APIs.
• Flexibility: Developers can integrate multiple models in their workflow based on cost, performance, and feature needs.
• Enhanced Productivity: Faster response times by utilizing optimized APIs from various providers.

---

Conclusion

By integrating API support for low-cost AI models, the Too Code VSCode extension will become a more powerful and flexible tool for developers. This will allow seamless AI assistance with cost-effective options while maintaining high performance.

Request Priority: High

Please consider adding this feature to enhance AI accessibility in VSCode. Looking forward to feedback from the community!