diff --git a/local/ai-commit/lua/ai-commit/spinner.lua b/local/ai-commit/lua/ai-commit/spinner.lua
index 9fd4f29..37692dc 100644
--- a/local/ai-commit/lua/ai-commit/spinner.lua
+++ b/local/ai-commit/lua/ai-commit/spinner.lua
@@ -197,6 +197,12 @@ function M.append_stream(spinner, text)
   if not spinner or not text or text == '' then
     return
   end
+  if vim.in_fast_event() then
+    vim.schedule(function()
+      M.append_stream(spinner, text)
+    end)
+    return
+  end
 
   local chunks = vim.split(text, '\n', { plain = true })
   for index, chunk in ipairs(chunks) do
diff --git a/local/ai-provider/lua/ai-provider/curl.lua b/local/ai-provider/lua/ai-provider/curl.lua
index 7a39954..7d6f151 100644
--- a/local/ai-provider/lua/ai-provider/curl.lua
+++ b/local/ai-provider/lua/ai-provider/curl.lua
@@ -100,7 +100,7 @@ end
 ---@field timeout? integer Curl max-time in milliseconds.
 ---@field is_cancelled? fun(): boolean Optional cancellation predicate.
 ---@field on_json_line? fun(data: table, line: string) Called for each decoded JSON line.
----@field callback fun(code: integer, error: string|nil) Called when curl exits.
+---@field callback fun(code: integer, error: string|nil, status: integer|nil) Called when curl exits.
 
 ---Run a streaming curl request and decode each stdout line as JSON.
 ---@param request AiProviderCurlStreamRequest
@@ -110,6 +110,7 @@ function M.stream_json_lines(request)
   local body = encode_body(request.body)
   local method = request.method or 'POST'
   local timeout = request.timeout or DEFAULT_TIMEOUT
+  local status_marker = '__AI_PROVIDER_HTTP_STATUS__:'
   local args = {
     '--silent',
     '--show-error',
@@ -118,6 +119,8 @@ function M.stream_json_lines(request)
     tostring(math.ceil(timeout / 1000)),
     '--request',
     method,
+    '--write-out',
+    '\n' .. status_marker .. '%{http_code}',
   }
 
   for name, value in pairs(json_headers(request.headers)) do
@@ -133,21 +136,43 @@ function M.stream_json_lines(request)
   table.insert(args, request.url)
 
   local stderr = {}
+  local stdout_lines = {}
+  local http_status = nil
   local job = Job:new {
     command = 'curl',
     args = args,
-    on_stdout = function(_, line)
-      if not line or line == '' or (request.is_cancelled and request.is_cancelled()) then
+    on_stdout = function(_, output)
+      if not output or output == '' or (request.is_cancelled and request.is_cancelled()) then
         return
       end
 
-      local ok, data = pcall(vim.json.decode, line)
-      if not ok or type(data) ~= 'table' then
-        return
-      end
-
-      if request.on_json_line then
-        request.on_json_line(data, line)
+      for line in tostring(output):gmatch('[^\r\n]+') do
+        line = line:gsub('^%s+', ''):gsub('%s+$', '')
+        if line ~= '' then
+          local status = line:match('^' .. status_marker .. '(%d+)$')
+          if status then
+            http_status = tonumber(status)
+          else
+            local payload = line
+            local event_data = line:match '^data:%s*(.*)$'
+            if event_data then
+              if event_data == '[DONE]' then
+                goto continue
+              end
+              payload = event_data
+            end
+
+            if #stdout_lines < 20 then
+              table.insert(stdout_lines, payload)
+            end
+
+            local ok, data = pcall(vim.json.decode, payload)
+            if ok and type(data) == 'table' and request.on_json_line then
+              schedule(request.on_json_line, data, line)
+            end
+          end
+        end
+        ::continue::
       end
     end,
     on_stderr = function(_, line)
@@ -156,7 +181,14 @@ function M.stream_json_lines(request)
       end
     end,
     on_exit = function(_, code)
-      schedule(request.callback, code, #stderr > 0 and table.concat(stderr, '\n') or nil)
+      local error_parts = {}
+      if #stderr > 0 then
+        table.insert(error_parts, table.concat(stderr, '\n'))
+      end
+      if http_status and http_status >= 400 and #stdout_lines > 0 then
+        table.insert(error_parts, table.concat(stdout_lines, '\n'))
+      end
+      schedule(request.callback, code, #error_parts > 0 and table.concat(error_parts, '\n') or nil, http_status)
     end,
   }
 
diff --git a/local/ai-provider/lua/ai-provider/providers/copilot.lua b/local/ai-provider/lua/ai-provider/providers/copilot.lua
index fe2715a..fc0135a 100644
--- a/local/ai-provider/lua/ai-provider/providers/copilot.lua
+++ b/local/ai-provider/lua/ai-provider/providers/copilot.lua
@@ -97,16 +97,20 @@ local function register_job(request, job)
   end
 end
 
+local function elapsed_ms_since(started_at)
+  return (vim.uv.hrtime() - started_at) / 1e6
+end
+
 local function get_api_token(request, callback)
   if state.api_token and state.api_token.expires_at and state.api_token.expires_at > os.time() then
     callback(state.api_token.token)
-    return
+    return nil
   end
 
   local oauth_token = get_oauth_token()
   if not oauth_token then
     callback(nil)
-    return
+    return nil
   end
 
   local job = curl.json {
@@ -138,6 +142,7 @@ local function get_api_token(request, callback)
     end,
   }
   register_job(request, job)
+  return job
 end
 
 function M.check(callback, opts)
@@ -187,6 +192,15 @@ end
 
 function M.chat(request)
   local started_at = vim.uv.hrtime()
+  local active_job = nil
+  local proxy_job = {
+    shutdown = function()
+      if active_job and active_job.shutdown then
+        active_job:shutdown()
+      end
+    end,
+  }
+  register_job(request, proxy_job)
 
   local function emit_status(phase, message)
     if request.on_status then
@@ -195,13 +209,13 @@ function M.chat(request)
         phase = phase,
         message = message,
         model = request.model,
-        elapsed_ms = (vim.uv.hrtime() - started_at) / 1e6,
+        elapsed_ms = elapsed_ms_since(started_at),
       }
     end
   end
 
   emit_status('authenticating', 'Authenticating with Copilot')
-  get_api_token(request, function(token)
+  local auth_job = get_api_token(request, function(token)
     if is_cancelled(request) then
       return
     end
@@ -211,7 +225,7 @@ function M.chat(request)
         request.callback(nil, {
           requested_model = request.model,
           used_model = request.model,
-          elapsed_ms = (vim.uv.hrtime() - started_at) / 1e6,
+          elapsed_ms = elapsed_ms_since(started_at),
           error = 'copilot authentication failed',
         })
       end
@@ -224,12 +238,14 @@ function M.chat(request)
     headers['Content-Type'] = 'application/json'
 
     local requested_model = request.model or AUTO_MODEL
-    local active_job = nil
 
     local function send_chat(model, retried_auto)
+      local chunks = {}
+      local used_model = model or AUTO_MODEL
+      local stream_error = nil
       local body = {
         messages = { { role = 'user', content = request.prompt } },
-        stream = false,
+        stream = request.stream ~= false,
         max_tokens = request.max_tokens,
       }
       if model and model ~= AUTO_MODEL then
@@ -237,6 +253,91 @@ function M.chat(request)
       end
 
       emit_status('generating', 'Generating response')
+      if body.stream then
+        active_job = curl.stream_json_lines {
+          method = 'POST',
+          url = endpoint .. '/chat/completions',
+          headers = headers,
+          body = body,
+          timeout = request.timeout or 30000,
+          is_cancelled = request.is_cancelled,
+          on_json_line = function(data)
+            if is_cancelled(request) then
+              return
+            end
+
+            used_model = data.model or used_model
+            if data.error then
+              stream_error = data.error
+            end
+            local choice = data.choices and data.choices[1]
+            local delta = choice and choice.delta and choice.delta.content
+            if type(delta) ~= 'string' or delta == '' then
+              return
+            end
+
+            table.insert(chunks, delta)
+            if request.on_chunk then
+              request.on_chunk(delta, data, 'message')
+            end
+          end,
+          callback = function(code, error_message, status)
+            if is_cancelled(request) then
+              return
+            end
+
+            local elapsed_ms = elapsed_ms_since(started_at)
+            if code ~= 0 or (status and status >= 400) or stream_error then
+              local error_message_with_status = 'copilot api request failed: ' .. tostring(status or code)
+              local error_code = type(stream_error) == 'table' and stream_error.code or nil
+              local error_body = error_message or (stream_error and vim.json.encode(stream_error)) or nil
+              log.error(string.format('%s elapsed_ms=%.0f body=%s', error_message_with_status, elapsed_ms, format_body_for_log(error_body)))
+              if status == 400 and (error_code == 'unsupported_api_for_model' or (error_body and error_body:match 'unsupported_api_for_model')) and body.model and not retried_auto then
+                log.warn('copilot model unsupported by streaming chat completions, retrying with auto: ' .. tostring(body.model))
+                send_chat(AUTO_MODEL, true)
+                return
+              end
+
+              if request.callback then
+                request.callback(nil, {
+                  requested_model = requested_model,
+                  used_model = used_model,
+                  elapsed_ms = elapsed_ms,
+                  error = error_message_with_status,
+                })
+              end
+              return
+            end
+
+            local message = table.concat(chunks, '')
+            if message == '' then
+              log.error('copilot streaming response missing message')
+              if request.callback then
+                request.callback(nil, {
+                  requested_model = requested_model,
+                  used_model = used_model,
+                  elapsed_ms = elapsed_ms,
+                  error = 'copilot returned no content',
+                })
+              end
+              return
+            end
+
+            log.info(string.format('copilot streaming response requested_model=%s used_model=%s elapsed_ms=%.0f', requested_model, used_model, elapsed_ms))
+            emit_status('done', 'Response complete')
+            if request.callback then
+              request.callback(message, {
+                requested_model = requested_model,
+                used_model = used_model,
+                elapsed_ms = elapsed_ms,
+              })
+            end
+          end,
+        }
+        register_job(request, active_job)
+        return
+      end
+
       active_job = curl.json {
         method = 'POST',
         url = endpoint .. '/chat/completions',
@@ -248,7 +349,7 @@ function M.chat(request)
             return
           end
 
-          local elapsed_ms = (vim.uv.hrtime() - started_at) / 1e6
+          local elapsed_ms = elapsed_ms_since(started_at)
           if response.status ~= 200 then
             local error_code = response.json and response.json.error and response.json.error.code
             if response.status == 400 and error_code == 'unsupported_api_for_model' and body.model and not retried_auto then
@@ -301,8 +402,11 @@ function M.chat(request)
     end
 
     send_chat(request.model or AUTO_MODEL, false)
-    return active_job
   end)
+  if auth_job then
+    active_job = auth_job
+  end
+  return proxy_job
 end
 
 return M
diff --git a/local/ai-provider/lua/ai-provider/providers/ollama.lua b/local/ai-provider/lua/ai-provider/providers/ollama.lua
index d20d084..88d623d 100644
--- a/local/ai-provider/lua/ai-provider/providers/ollama.lua
+++ b/local/ai-provider/lua/ai-provider/providers/ollama.lua
@@ -392,24 +392,25 @@ function M.chat(request)
           queue_chunk(chunk, data, 'message')
         end
       end,
-      callback = function(code, error_message)
+      callback = function(code, error_message, status)
         if is_cancelled(request) then
           return
         end
 
-        if code ~= 0 then
-          log.error('ollama request process failed code=' .. tostring(code) .. ' model=' .. tostring(raw_model) .. ' error=' .. tostring(error_message))
+        if code ~= 0 or (status and status >= 400) then
+          local error_detail = error_message or (status and ('HTTP ' .. status) or 'ollama request failed')
+          log.error('ollama request process failed code=' .. tostring(code) .. ' status=' .. tostring(status) .. ' model=' .. tostring(raw_model) .. ' error=' .. tostring(error_detail))
           if request.callback then
             request.callback(nil, {
               requested_model = selected_model,
               used_model = final_model,
               elapsed_ms = (vim.uv.hrtime() - started_at) / 1e6,
-              error = error_message or 'ollama request failed',
+              error = error_detail,
             })
           end
           emit_status {
             phase = 'error',
-            message = error_message or 'Ollama request failed',
+            message = error_detail,
           }
           return
         end
@@ -536,6 +537,10 @@ function M.chat(request)
         phase = 'loading',
         message = 'Loading model',
       }
+      emit_status {
+        phase = 'context',
+        message = 'Loading prompt context',
+      }
     else
       emit_status {
         phase = 'context',
diff --git a/local/ai-provider/tests/ai_provider/copilot_spec.lua b/local/ai-provider/tests/ai_provider/copilot_spec.lua
new file mode 100644
index 0000000..8af3750
--- /dev/null
+++ b/local/ai-provider/tests/ai_provider/copilot_spec.lua
@@ -0,0 +1,94 @@
+describe('copilot provider', function()
+  local original_curl
+  local original_path
+  local original_log
+
+  before_each(function()
+    original_curl = package.loaded['ai-provider.curl']
+    original_path = package.loaded['plenary.path']
+    original_log = package.loaded['ai-provider.log']
+    package.loaded['ai-provider.providers.copilot'] = nil
+    package.loaded['ai-provider.log'] = {
+      debug = function() end,
+      info = function() end,
+      warn = function() end,
+      error = function() end,
+    }
+
+    local fake_path = {}
+    function fake_path:joinpath()
+      return self
+    end
+    function fake_path:exists()
+      return true
+    end
+    function fake_path:read()
+      return vim.json.encode { ['github.com'] = { oauth_token = 'oauth-token' } }
+    end
+
+    package.loaded['plenary.path'] = {
+      new = function()
+        return fake_path
+      end,
+    }
+  end)
+
+  after_each(function()
+    package.loaded['ai-provider.curl'] = original_curl
+    package.loaded['plenary.path'] = original_path
+    package.loaded['ai-provider.log'] = original_log
+    package.loaded['ai-provider.providers.copilot'] = nil
+  end)
+
+  it('streams chat completion deltas through provider chunks', function()
+    local captured_body = nil
+    package.loaded['ai-provider.curl'] = {
+      json = function(request)
+        assert.matches('/copilot_internal/v2/token$', request.url)
+        request.callback {
+          status = 200,
+          json = {
+            token = 'api-token',
+            expires_at = os.time() + 60,
+            endpoints = { api = 'https://copilot.example.test' },
+          },
+        }
+        return { shutdown = function() end }
+      end,
+      stream_json_lines = function(request)
+        captured_body = request.body
+        request.on_json_line { model = 'gpt-4o', choices = { { delta = { content = 'he' } } } }
+        request.on_json_line { model = 'gpt-4o', choices = { { delta = { content = 'llo' } } } }
+        request.callback(0, nil, 200)
+        return { shutdown = function() end }
+      end,
+    }
+
+    local copilot = require 'ai-provider.providers.copilot'
+    local chunks = {}
+    local message = nil
+    local meta = nil
+
+    local job = copilot.chat {
+      model = 'gpt-4o',
+      prompt = 'Say hello',
+      on_chunk = function(chunk)
+        table.insert(chunks, chunk)
+      end,
+      callback = function(result, result_meta)
+        message = result
+        meta = result_meta
+      end,
+    }
+
+    assert.is_table(job)
+    assert.is_table(captured_body)
+    ---@cast captured_body table
+    assert.are.same(true, captured_body.stream)
+    assert.are.same('hello', table.concat(chunks, ''))
+    assert.are.same('hello', message)
+    assert.is_table(meta)
+    ---@cast meta table
+    assert.are.same('gpt-4o', meta.used_model)
+  end)
+end)
diff --git a/local/ai-provider/tests/ai_provider/ollama_options_spec.lua b/local/ai-provider/tests/ai_provider/ollama_options_spec.lua
index 4783e92..4cc3225 100644
--- a/local/ai-provider/tests/ai_provider/ollama_options_spec.lua
+++ b/local/ai-provider/tests/ai_provider/ollama_options_spec.lua
@@ -318,11 +318,11 @@ describe('ollama provider options', function()
     }
 
     vim.wait(1000, function()
-      return #statuses >= 3
+      return #statuses >= 4
     end, 10)
 
     assert.are.same('loading', statuses[1])
-    assert.are.same('generating', statuses[2])
-    assert.is_false(vim.tbl_contains(statuses, 'context'))
+    assert.are.same('context', statuses[2])
+    assert.are.same('generating', statuses[3])
   end)
 end)