add API to get a list of all available models

[arunabhcode]

Addressing #828

It adds two calls listModels() and listProviders().

[CLAassistant]

All committers have signed the CLA.

[arunabhcode]

Test Code:

import AI from './AI.js';

// Create test instance
const ai = new AI({
    authToken: 'test-token',
    APIOrigin: 'test-origin',
    appID: 'test-app'
});

// Test listModels
const models = ai.listModels();
const providers = ai.listProviders();
console.log('Available models:', JSON.stringify(models, null, 2));
console.log('Available providers:', JSON.stringify(providers, null, 2));

Sample output:

Available models: {
  "chat": {
    "openai": [
      "gpt-4o",
      "gpt-4o-mini"
    ],
    "anthropic": [
      "claude-3-haiku-20240307",
      "claude-3-5-sonnet-20240620",
      "claude-3-5-sonnet-20241022",
      "claude-3-5-sonnet-latest",
      "claude-3-7-sonnet-latest"
    ],
    "together": [
      "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
      "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
      "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo",
      "google/gemma-2-27b-it"
    ],
    "mistral": [
      "mistral-large-latest",
      "codestral-latest"
    ],
    "groq": [
      "distil-whisper-large-v3-en",
      "gemma2-9b-it",
      "gemma-7b-it",
      "llama-3.1-70b-versatile",
      "llama-3.1-8b-instant",
      "llama3-70b-8192",
      "llama3-8b-8192",
      "llama3-groq-70b-8192-tool-use-preview",
      "llama3-groq-8b-8192-tool-use-preview",
      "llama-guard-3-8b",
      "mixtral-8x7b-32768",
      "whisper-large-v3"
    ],
    "xai": [
      "grok-beta"
    ],
    "deepseek": [
      "deepseek-chat",
      "deepseek-reasoner"
    ],
    "gemini": [
      "gemini-1.5-flash",
      "gemini-2.0-flash"
    ]
  }
}
Available providers: {
  "chat": [
    "openai",
    "anthropic",
    "together",
    "mistral",
    "groq",
    "xai",
    "deepseek",
    "gemini"
  ]
}

[KernelDeimos]

Hi, thanks for the contribution. This unfortunately doesn't address the issue properly - that's maybe on us because the issue you linked to doesn't have any description. I'll add further detail on monday. This should be requesting model information from the backend services in the AI module. I'll try to remember to add more information about this by the end of Monday.

[arunabhcode]

@KernelDeimos Sounds good, waiting for further clarification will try to work on it in the meantime and guess around what's needed.

[KernelDeimos]

Hi again, this request to backend gets all supported models

await puter.drivers.call('puter-chat-completion', 'ai-chat', 'models');

This endpoint doesn't split it into "models" and "providers", but it does provide enough information for you to derive that.

[arunabhcode]

@KernelDeimos so you want to retrieve the dict returned by calling this and then process that dict to create the dict that I hardcoded?

[KernelDeimos]

Yes that's correct

[arunabhcode]

@avijh squashed our commits into one since this is a small PR.

[KernelDeimos]

Suggested change

	if (item.provider) {
	if (item.provider) {

[KernelDeimos]

I think this is supposed to be intended one more level

Suggested change

	if (item.provider) {
	providers.add(item.provider);
	}
	if (item.provider) {
	providers.add(item.provider);
	}

[KernelDeimos]

Whenever you see if if if if, you can be sure there's a simpler approach. This amount of nesting increases a metric in static code analysis known as "cognitive complexity" which is a measure of how difficult the code is for somebody to understand and maintain later.

We can use guard clauses (also I found this video which is a bit long but you can get the jist of it by just scanning through real quick and looking at how he's changing the code) to simplify (untested):

...forEach(item => {
    if ( ! item.id ) return;
    if ( provider && ( item.provider !== provider ) ) return;
    if ( ! modelsByProvider[item.provider] ) {
      modelsByProvider[item.provider] = [];
    }
    modelsByProvider[item.provider].push(item.id);
}

Technically we could even combine the first two if statements, but I think it's easier to read when separate concerns still belong on separate lines.

[avijh]

Agreed. I modified the code per your suggestions and also removed redundant {} for readability.

[KernelDeimos]

Manual test passed. I did manual testing early since I know testing this without API keys is difficult. I'll do a quick test again after my review comments are addressed, and then this will be ready to merge.

pr-1160_1.7x.mp4

[KernelDeimos]

Nice work! Thanks for addressing my comments