feat: add AI Stupid Level provider with 6 auto-routing models

[StudioPlatforms]

Summary

This PR adds AI Stupid Level integration to Continue, providing 6 intelligent auto-routing models that automatically select the best AI model for each task.

Changes Made

1. core/llm/llms/AIStupidLevel.ts (New File)

• Created new AI Stupid Level provider extending OpenAI base class
• Configured with custom baseURL (https://aistupidlevel.info/v1/)
• Set default model to auto-coding for optimal coding performance

2. core/llm/llms/index.ts

• Added import for AIStupidLevel provider
• Registered AIStupidLevel in LLMClasses array for provider discovery

Available Models

Users can configure any of these 6 auto-routing models:

• auto - Best overall performance
• auto-coding - Optimized for coding tasks (default)
• auto-reasoning - Optimized for reasoning tasks
• auto-creative - Optimized for creative tasks
• auto-fastest - Optimized for speed
• auto-cheapest - Optimized for cost efficiency

Benefits

• Intelligent Routing: Automatically selects the best model based on task requirements
• Multi-Provider Support: Leverages multiple AI providers (Anthropic, OpenAI, Google, etc.)
• Cost Optimization: Choose between performance, speed, or cost
• Easy Integration: Works seamlessly with existing Continue architecture

Configuration Example

{
  "models": [
    {
      "provider": "aistupidlevel",
      "model": "auto-coding",
      "apiKey": "your_api_key_here"
    }
  ]
}

Environment Variable

AI_STUPID_LEVEL_API_KEY=your_api_key_here

Users can obtain an API key from https://aistupidlevel.info

---

Summary by cubic

Adds Stupid Level provider that auto-routes to the best model per task. Registers it for discovery and sets auto-coding as the default.

• New Features
  • Supports models: auto, auto-coding (default), auto-reasoning, auto-creative, auto-fastest, auto-cheapest.

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[StudioPlatforms]

I have read the CLA Document and I hereby sign the CLA

[cubic-dev-ai]

No issues found across 2 files

[bdougie]

@StudioPlatforms thanks for opening this PR. Just want to point that we appreciate folks adding new providers, but it would also require a docs update as well. https://docs.continue.dev/customize/model-providers/overview

[StudioPlatforms]

@bdougie Thank you for the feedback! You're absolutely right - apologies for missing that initially. I've just pushed comprehensive documentation updates:

✅ Added detailed provider documentation at docs/customize/model-providers/more/aistupidlevel.mdx
✅ Updated the model providers overview to include AIStupidLevel in the Hosted Services section

The documentation includes:

• Overview of what AIStupidLevel is (intelligent AI router with real-time benchmarks)
• Configuration examples in both YAML and JSON formats
• Detailed explanation of all 6 routing strategies (auto, auto-coding, auto-reasoning, auto-creative, auto-cheapest, auto-fastest)
• How it works section with setup steps
• Key features and benefits
• Response headers information
• Links to the website and community resources

Please let me know if there's anything else needed for the documentation or if you'd like any adjustments!

[cubic-dev-ai]

No issues found across 2 files

[bdougie]

After chatting with the team internally @StudioPlatforms. We welcome contributions from providers where there is a clear need. Feel free to correct me, but it doesn't feel like there is a lot of demand for the stupid level provider in market.

[StudioPlatforms]

After chatting with the team internally @StudioPlatforms. We welcome contributions from providers where there is a clear need. Feel free to correct me, but it doesn't feel like there is a lot of demand for the stupid level provider in market.

Thanks for the thoughtful feedback, @bdougie and the Cline team! A quick data-point dump on demand/traction for the AI Stupid Level provider so you can judge whether it’s worth having as a community provider (i’m happy to maintain it end-to-end).

We have press & coverage: Notebookcheck wrote about AI Stupid Level as a real-time benchmarking/router tool tracking top models, calling it “first-of-its-kind”. https://www.notebookcheck.net/Open-source-tool-measures-the-stupidity-level-of-AI-models-1117315.0.html

Another mention in Gizmochina covered it as an open-source tool monitoring LLM drift in real time. https://www.gizmochina.com/2025/09/18/open-source-tool-now-measures-the-stupidity-level-of-ai-models-in-real-time/

On Reddit there’s a dedicated community https://www.reddit.com/r/AIStupidLevel with release and update discussion.

Also aistupidlevel.info has passed 1 million visits in under 2 months of existence.

Furthermore i can point to YouTube, Instagram, TikTok and LinkedIn posts from users and creators sharing results, demos and routing-integration work (i can send direct links if helpful). For example, the “Pro TV” feature (where we got a segment) shows the demand outside dev circles.

If you guys really don’t consider it something worth including right now, no hard feelings at all we can close the PR.

[bdougie]

If you guys really don’t consider it something worth including right now, no hard feelings at all we can close the PR.

Closing for now. I think you are doing a great job selling, but it would be nice to get users to come to this consensus.