diff --git a/local/ai-provider/lua/ai-provider/init.lua b/local/ai-provider/lua/ai-provider/init.lua
index 1b91665..7cb9e95 100644
--- a/local/ai-provider/lua/ai-provider/init.lua
+++ b/local/ai-provider/lua/ai-provider/init.lua
@@ -51,6 +51,7 @@
 ---@field model? string Model/profile name.
 ---@field used_model? string Raw provider model name.
 ---@field tokens? integer Token count when the provider reports one.
+---@field tokens_per_second? number Generation throughput when the provider reports enough timing data.
 ---@field elapsed_ms? number Elapsed duration in milliseconds.
 
 ---@class AiProviderChatRequest
diff --git a/local/ai-provider/lua/ai-provider/providers/ollama.lua b/local/ai-provider/lua/ai-provider/providers/ollama.lua
index b29357a..3be98f8 100644
--- a/local/ai-provider/lua/ai-provider/providers/ollama.lua
+++ b/local/ai-provider/lua/ai-provider/providers/ollama.lua
@@ -136,6 +136,7 @@ function M.chat(request)
   local metrics = {}
   local thinking_chars = 0
   local last_status_key = nil
+  local generation_started_at = nil
 
   local function emit_status(status)
     if not request.on_status then
@@ -146,7 +147,7 @@ function M.chat(request)
     status.model = status.model or selected_model
     status.used_model = status.used_model or final_model
     status.elapsed_ms = status.elapsed_ms or elapsed_ms_since(started_at)
-    local key = table.concat({ status.phase or '', status.message or '', tostring(status.tokens), tostring(status.used_model) }, '|')
+    local key = table.concat({ status.phase or '', status.message or '', tostring(status.tokens_per_second), tostring(status.used_model) }, '|')
     if key == last_status_key then
       return
     end
@@ -207,13 +208,20 @@ function M.chat(request)
         metrics.prompt_eval_duration = data.prompt_eval_duration or metrics.prompt_eval_duration
         metrics.eval_count = data.eval_count or metrics.eval_count
         metrics.eval_duration = data.eval_duration or metrics.eval_duration
+        if not generation_started_at and data.eval_count then
+          generation_started_at = vim.uv.hrtime()
+        end
+        local status_tokens_per_second = tokens_per_second(data.eval_count, data.eval_duration)
+        if not status_tokens_per_second and generation_started_at and data.eval_count and data.eval_count > 0 then
+          status_tokens_per_second = data.eval_count / ((vim.uv.hrtime() - generation_started_at) / 1e9)
+        end
         local thinking = data.message and data.message.thinking or ''
         if thinking ~= '' then
           thinking_chars = thinking_chars + #thinking
           emit_status {
             phase = 'thinking',
             message = 'Thinking',
-            tokens = data.eval_count,
+            tokens_per_second = status_tokens_per_second,
           }
         end
         local chunk = data.message and data.message.content or ''
@@ -222,7 +230,7 @@ function M.chat(request)
           emit_status {
             phase = thinking_chars > 0 and 'generating' or 'generating',
             message = 'Generating response',
-            tokens = data.eval_count,
+            tokens_per_second = status_tokens_per_second,
           }
           if request.on_chunk then
             vim.schedule(function()
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 e4bbd81..4aff6a6 100644
--- a/local/ai-provider/tests/ai_provider/ollama_options_spec.lua
+++ b/local/ai-provider/tests/ai_provider/ollama_options_spec.lua
@@ -168,9 +168,9 @@ describe('ollama provider options', function()
   it('emits standardized thinking and generating status events', function()
     package.loaded['ai-provider.curl'] = {
       stream_json_lines = function(request)
-        request.on_json_line { model = 'gemma4:e2b', message = { thinking = 'thinking...' }, eval_count = 7 }
-        request.on_json_line { model = 'gemma4:e2b', message = { content = 'ok' }, eval_count = 9 }
-        request.on_json_line { model = 'gemma4:e2b', done_reason = 'stop', eval_count = 9 }
+        request.on_json_line { model = 'gemma4:e2b', message = { thinking = 'thinking...' }, eval_count = 7, eval_duration = 1000000000 }
+        request.on_json_line { model = 'gemma4:e2b', message = { content = 'ok' }, eval_count = 9, eval_duration = 1000000000 }
+        request.on_json_line { model = 'gemma4:e2b', done_reason = 'stop', eval_count = 9, eval_duration = 1000000000 }
         request.callback(0)
         return { shutdown = function() end }
       end,
@@ -195,9 +195,9 @@ describe('ollama provider options', function()
 
     assert.are.same('generating', statuses[1].phase)
     assert.are.same('thinking', statuses[2].phase)
-    assert.are.same(7, statuses[2].tokens)
+    assert.are.same(7, statuses[2].tokens_per_second)
     assert.are.same('generating', statuses[3].phase)
-    assert.are.same(9, statuses[3].tokens)
+    assert.are.same(9, statuses[3].tokens_per_second)
     assert.are.same('done', statuses[4].phase)
   end)
 end)
diff --git a/lua/plugins/ai-commit.lua b/lua/plugins/ai-commit.lua
index 324bc63..114557e 100644
--- a/lua/plugins/ai-commit.lua
+++ b/lua/plugins/ai-commit.lua
@@ -265,11 +265,11 @@ local function complete_ollama(full_prompt, callback, status_callback, request_c
     elseif status.phase == 'loaded' then
       status_callback('Loaded model ' .. status_model)
     elseif status.phase == 'thinking' then
-      local token_suffix = status.tokens and (' (' .. status.tokens .. ' tokens)') or ''
-      status_callback('Thinking with ' .. status_model .. token_suffix)
+      local speed_suffix = status.tokens_per_second and string.format(' (%.1f t/s)', status.tokens_per_second) or ''
+      status_callback('Thinking with ' .. status_model .. speed_suffix)
     elseif status.phase == 'generating' then
-      local token_suffix = status.tokens and (' (' .. status.tokens .. ' tokens)') or ''
-      status_callback('Generating response with ' .. status_model .. token_suffix)
+      local speed_suffix = status.tokens_per_second and string.format(' (%.1f t/s)', status.tokens_per_second) or ''
+      status_callback('Generating response with ' .. status_model .. speed_suffix)
     elseif status.phase == 'error' then
       status_callback('Provider error from ' .. status_model)
     end