Facing the same issue. Is there a work around i can use till the fix is propagated to release?

@rajasrijan

There is a way to temporarily fix this issue:

• Locate your VSCode extensions directory. For the official VSCode version, it is located at: C:\Users\<your username>\.vscode\extensions
• Back up the file C:\Users\<your username>\.vscode\extensions\github.copilot-chat-0.37.2\dist\extension.js
• Open this file
• Search for "getAllModels", find the last result, and locate this section:

async getAllModels(t,r,a){if(!a)return[];let o=a.url;try{await this._checkOllamaVersion(o);let c=(await(await this._fetcherService.fetch(`${o}/api/tags`,{method:"GET"})).json()).models,l={};for(let d of c){let u=this._modelCache.get(`${o}/${d.model}`);u||(u=await this._getOllamaModelInfo(o,d.model),this._modelCache.set(`${o}/${d.model}`,u)),l[d.model]={maxInputTokens:u.capabilities.limits?.max_prompt_tokens??4096,maxOutputTokens:u.capabilities.limits?.max_output_tokens??4096,name:u.name,toolCalling:!!u.capabilities.supports.tool_calls,vision:!!u.capabilities.supports.vision}}return kP(this._name,c).map(d=>({...d,url:o}))}catch(s){throw s instanceof Error&&s.message.includes("Ollama server version")?s:new Error('Failed to fetch models from Ollama. Please ensure Ollama is running. If ollama is on another host, please configure the `"github.copilot.chat.byok.ollamaEndpoint"` setting.')}}

• Replace it with:

async getAllModels(t,r,a){if(!a)return[];let o=a.url;try{await this._checkOllamaVersion(o);let c=(await(await this._fetcherService.fetch(`${o}/api/tags`,{method:"GET"})).json()).models;this._knownModels={};for(let d of c){let u=this._modelCache.get(`${o}/${d.model}`);u||(u=await this._getOllamaModelInfo(o,d.model),this._modelCache.set(`${o}/${d.model}`,u)),this._knownModels[u.id]={maxInputTokens:u.capabilities.limits?.max_prompt_tokens??4096,maxOutputTokens:u.capabilities.limits?.max_output_tokens??4096,name:u.name,toolCalling:!!u.capabilities.supports.tool_calls,vision:!!u.capabilities.supports.vision}};return kP(this._name,this._knownModels).map(d=>({...d,url:o}))}catch(s){throw s instanceof Error&&s.message.includes("Ollama server version")?s:new Error('Failed to fetch models from Ollama. Please ensure Ollama is running. If ollama is on another host, please configure the `"github.copilot.chat.byok.ollamaEndpoint"` setting.')}}

Differences in the modified section:

• Save the file and restart VS Code.

@rajasrijan

There is a way to temporarily fix this issue:

* Locate your VSCode extensions directory. For the official VSCode version, it is located at: `C:\Users\<your username>\.vscode\extensions`

* Back up the file `C:\Users\<your username>\.vscode\extensions\github.copilot-chat-0.37.2\dist\extension.js`

* Open this file

* Search for "getAllModels", find the last result, and locate this section:

async getAllModels(t,r,a){if(!a)return[];let o=a.url;try{await this._checkOllamaVersion(o);let c=(await(await this._fetcherService.fetch(`${o}/api/tags`,{method:"GET"})).json()).models,l={};for(let d of c){let u=this._modelCache.get(`${o}/${d.model}`);u||(u=await this._getOllamaModelInfo(o,d.model),this._modelCache.set(`${o}/${d.model}`,u)),l[d.model]={maxInputTokens:u.capabilities.limits?.max_prompt_tokens??4096,maxOutputTokens:u.capabilities.limits?.max_output_tokens??4096,name:u.name,toolCalling:!!u.capabilities.supports.tool_calls,vision:!!u.capabilities.supports.vision}}return kP(this._name,c).map(d=>({...d,url:o}))}catch(s){throw s instanceof Error&&s.message.includes("Ollama server version")?s:new Error('Failed to fetch models from Ollama. Please ensure Ollama is running. If ollama is on another host, please configure the `"github.copilot.chat.byok.ollamaEndpoint"` setting.')}}

* Replace it with:

async getAllModels(t,r,a){if(!a)return[];let o=a.url;try{await this._checkOllamaVersion(o);let c=(await(await this._fetcherService.fetch(`${o}/api/tags`,{method:"GET"})).json()).models;this._knownModels={};for(let d of c){let u=this._modelCache.get(`${o}/${d.model}`);u||(u=await this._getOllamaModelInfo(o,d.model),this._modelCache.set(`${o}/${d.model}`,u)),this._knownModels[u.id]={maxInputTokens:u.capabilities.limits?.max_prompt_tokens??4096,maxOutputTokens:u.capabilities.limits?.max_output_tokens??4096,name:u.name,toolCalling:!!u.capabilities.supports.tool_calls,vision:!!u.capabilities.supports.vision}};return kP(this._name,this._knownModels).map(d=>({...d,url:o}))}catch(s){throw s instanceof Error&&s.message.includes("Ollama server version")?s:new Error('Failed to fetch models from Ollama. Please ensure Ollama is running. If ollama is on another host, please configure the `"github.copilot.chat.byok.ollamaEndpoint"` setting.')}}

Differences in the modified section:

* Save the file and restart VS Code.

That didn't work. I can see the models now with the capabilities. but cant use them anymore.

Sorry, your request failed. Please try again.

Copilot Request id: a1c18984-5dd7-450b-a333-e5a85639f7ad

Reason: 404 page not found: Error: 404 page not found at x3._provideLanguageModelResponse (/home/srijan/.vscode/extensions/github.copilot-chat-0.37.2/dist/extension.js:1409:11576) at process.processTicksAndRejections (node:internal/process/task_queues:105:5)

Tested PR in VS Code Insiders. Same as @rajasrijan, I can see the models but none of them work.

@wrenchpilot @rajasrijan

I confirmed the issue. Previously I was previously using a self-implemented Ollama service, so I didn’t notice this problem.
The issue is that Ollama’s endpoint is not /chat/completions, but /api/chat.
I will fix this issue.

@wrenchpilot @rajasrijan

The 404 issue was not caused by this PR, so I submitted a new one that should resolve it.

#3684

Should this be fixed? I'm running v0.37.8 with vscode 1.109.5 and I'm still experiencing this issue

EDIT: nevermind, sorry. I have manually removed extension files from .vscode/extensions and redownloaded it. It still returns 404, but models are visible.

I was still facing the same issue. The models were visible in ask mode but not in "plan" and "agent" mode.

Ask Mode:

Agent Mode:

Is it fixed? In 0.37.9 I'm still experiencing the issue. Even deleted and redownloaded the extension.
ollama version is 0.17.4

I think it got worse. It is showing less models under ask than before and nothing under agent still.