feat(copilot): Support "Auto" model mode (Smart Model Selection) similar to VS Code (#1518)

* update: add auto into model completions

* update: add auto model selection module

* fix: remove unnecessary notifications

* ui: show current model at header

* refactor: remove function to provider configs

* fix: rename duplicate function name

* fix: remove route_model provider to a better position

* fix: add the 'auto' model option as an model in copilot provider

* refactor: move model selector to specific provider

* cleanup some stuff

Signed-off-by: Tomas Slusny <slusnucky@gmail.com>

---------

Signed-off-by: Tomas Slusny <slusnucky@gmail.com>
Co-authored-by: Tomas Slusny <slusnucky@gmail.com>

Author: pxwg

lua/CopilotChat/client.lua

@@ -319,6 +319,16 @@ function Client:ask(opts)
     error('Provider not found: ' .. provider_name)
   end
 
+  if provider.resolve_model then
+    local headers = self:authenticate(provider_name)
+    local resolved_model = provider.resolve_model(headers, opts.model)
+    opts.model = resolved_model
+    model_config = models[opts.model]
+    if not model_config then
+      error('Resolved model not found: ' .. opts.model)
+    end
+  end
+
   local options = {
     model = vim.tbl_extend('force', model_config, {
       id = opts.model:gsub(':' .. provider_name .. '$', ''),
@@ -389,6 +399,7 @@ function Client:ask(opts)
   local errored = nil
   local finished = false
   local token_count = 0
+  local out_model = nil
   local response_content_buffer = stringbuffer()
   local response_reasoning_buffer = stringbuffer()
 
@@ -451,6 +462,10 @@ function Client:ask(opts)
       response_reasoning_buffer:put(out.reasoning)
     end
 
+    if out.model then
+      out_model = out.model
+    end
+
     if opts.on_progress then
       opts.on_progress({
         role = constants.ROLE.ASSISTANT,
@@ -589,6 +604,7 @@ function Client:ask(opts)
       content = response_text,
       reasoning = response_reasoning,
       tool_calls = #tool_calls:values() > 0 and tool_calls:values() or nil,
+      model = out_model,
     },
     token_count = token_count,
     token_max_count = max_tokens,

lua/CopilotChat/config/providers.lua

@@ -197,7 +197,7 @@ local function get_github_models_token(tag)
 end
 
 --- Prepare input for Responses API
----@param inputs table<CopilotChat.client.Message>
+---@param inputs CopilotChat.client.Message[]
 ---@param opts CopilotChat.config.providers.Options
 ---@return table
 local function prepare_responses_input(inputs, opts)
@@ -257,7 +257,7 @@ local function prepare_responses_input(inputs, opts)
 end
 
 --- Prepare input for Chat Completions API
----@param inputs table<CopilotChat.client.Message>
+---@param inputs CopilotChat.client.Message[]
 ---@param opts CopilotChat.config.providers.Options
 ---@return table
 local function prepare_chat_input(inputs, opts)
@@ -353,6 +353,7 @@ local function prepare_responses_output(output)
   local finish_reason = nil
   local total_tokens = nil
   local tool_calls = {}
+  local model = nil
 
   -- Handle errors
   local error_msg = output.error or (output.response and output.response.error)
@@ -366,6 +367,7 @@ local function prepare_responses_output(output)
       finish_reason = 'error: ' .. tostring(error_msg),
       total_tokens = nil,
       tool_calls = {},
+      model = nil,
     }
   end
 
@@ -398,6 +400,9 @@ local function prepare_responses_output(output)
         if response.usage then
           total_tokens = response.usage.total_tokens
         end
+        if response.model then
+          model = response.model
+        end
         finish_reason = 'stop'
       end
     elseif output.type == 'response.failed' then
@@ -429,6 +434,9 @@ local function prepare_responses_output(output)
     if response.usage then
       total_tokens = response.usage.total_tokens
     end
+    if response.model then
+      model = response.model
+    end
     finish_reason = response.status == 'completed' and 'stop' or nil
   end
 
@@ -438,6 +446,7 @@ local function prepare_responses_output(output)
     finish_reason = finish_reason,
     total_tokens = total_tokens,
     tool_calls = tool_calls,
+    model = model,
   }
 end
 
@@ -477,13 +486,15 @@ local function prepare_chat_output(output)
   local reasoning = message and (message.reasoning or message.reasoning_content)
   local usage = choice.usage and choice.usage.total_tokens or output.usage and output.usage.total_tokens
   local finish_reason = choice.finish_reason or choice.done_reason or output.finish_reason or output.done_reason
+  local model = choice.model or output.model
 
   return {
     content = content,
     reasoning = reasoning,
     finish_reason = finish_reason,
     total_tokens = usage,
     tool_calls = tool_calls,
+    model = model,
   }
 end
 
@@ -498,13 +509,15 @@ end
 ---@field finish_reason string?
 ---@field total_tokens number?
 ---@field tool_calls table<CopilotChat.client.ToolCall>
+---@field model string?
 
 ---@class CopilotChat.config.providers.Provider
 ---@field disabled nil|boolean
 ---@field get_headers nil|fun():table<string, string>,number?
 ---@field get_info nil|fun(headers:table):string[]
 ---@field get_models nil|fun(headers:table):table<CopilotChat.client.Model>
----@field prepare_input nil|fun(inputs:table<CopilotChat.client.Message>, opts:CopilotChat.config.providers.Options):table
+---@field resolve_model nil|fun(headers:table, model: string):string
+---@field prepare_input nil|fun(inputs:CopilotChat.client.Message[], opts:CopilotChat.config.providers.Options):table,table?
 ---@field prepare_output nil|fun(output:table, opts:CopilotChat.config.providers.Options):CopilotChat.config.providers.Output
 ---@field get_url nil|fun(opts:CopilotChat.config.providers.Options):string
 
@@ -529,6 +542,7 @@ M.copilot = {
       ['Editor-Version'] = EDITOR_VERSION,
       ['Editor-Plugin-Version'] = 'CopilotChat.nvim/*',
       ['Copilot-Integration-Id'] = 'vscode-chat',
+      ['x-github-api-version'] = '2025-10-01',
     },
       response.body.expires_at
   end,
@@ -637,9 +651,36 @@ M.copilot = {
       end
     end
 
+    -- Auto model selector
+    table.insert(models, {
+      id = 'auto',
+      name = 'Auto (Copilot)',
+      description = 'Auto selects the best model for your request.',
+    })
+
     return models
   end,
 
+  resolve_model = function(headers, model)
+    if model ~= 'auto' then
+      return model
+    end
+
+    local url = 'https://api.githubcopilot.com/models/session'
+    local response, err = curl.post(url, {
+      headers = headers,
+      body = { auto_mode = { model_hints = { 'auto' } } },
+      json_response = true,
+      json_request = true,
+    })
+
+    if err then
+      error(err)
+    end
+
+    return response.body.selected_model
+  end,
+
   prepare_input = function(inputs, opts)
     local request
     if opts.model.use_responses then

lua/CopilotChat/ui/chat.lua

@@ -771,6 +771,9 @@ function Chat:render()
       -- Overlay section header with nice display
       local header_value = self.headers[message.role]
       local header_line = message.section.start_line - 2
+      if message.model then
+        header_value = header_value .. ' (' .. message.model .. ')'
+      end
 
       vim.api.nvim_buf_set_extmark(self.bufnr, highlight_ns, header_line, 0, {
         conceal = '',