[pull] master from Comfy-Org:master

comfy/sample.py

@@ -64,10 +64,10 @@ def sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative
     sampler = comfy.samplers.KSampler(model, steps=steps, device=model.load_device, sampler=sampler_name, scheduler=scheduler, denoise=denoise, model_options=model.model_options)
 
     samples = sampler.sample(noise, positive, negative, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
-    samples = samples.to(comfy.model_management.intermediate_device())
+    samples = samples.to(device=comfy.model_management.intermediate_device(), dtype=comfy.model_management.intermediate_dtype())
     return samples
 
 def sample_custom(model, noise, cfg, sampler, sigmas, positive, negative, latent_image, noise_mask=None, callback=None, disable_pbar=False, seed=None):
     samples = comfy.samplers.sample(model, noise, positive, negative, cfg, model.load_device, sampler, sigmas, model_options=model.model_options, latent_image=latent_image, denoise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
-    samples = samples.to(comfy.model_management.intermediate_device())
+    samples = samples.to(device=comfy.model_management.intermediate_device(), dtype=comfy.model_management.intermediate_dtype())
     return samples

comfy/sd1_clip.py

@@ -46,7 +46,7 @@ def encode_token_weights(self, token_weight_pairs):
         out, pooled = o[:2]
 
         if pooled is not None:
-            first_pooled = pooled[0:1].to(model_management.intermediate_device())
+            first_pooled = pooled[0:1].to(device=model_management.intermediate_device(), dtype=model_management.intermediate_dtype())
         else:
             first_pooled = pooled
 
@@ -63,16 +63,16 @@ def encode_token_weights(self, token_weight_pairs):
             output.append(z)
 
         if (len(output) == 0):
-            r = (out[-1:].to(model_management.intermediate_device()), first_pooled)
+            r = (out[-1:].to(device=model_management.intermediate_device(), dtype=model_management.intermediate_dtype()), first_pooled)
         else:
-            r = (torch.cat(output, dim=-2).to(model_management.intermediate_device()), first_pooled)
+            r = (torch.cat(output, dim=-2).to(device=model_management.intermediate_device(), dtype=model_management.intermediate_dtype()), first_pooled)
 
         if len(o) > 2:
             extra = {}
             for k in o[2]:
                 v = o[2][k]
                 if k == "attention_mask":
-                    v = v[:sections].flatten().unsqueeze(dim=0).to(model_management.intermediate_device())
+                    v = v[:sections].flatten().unsqueeze(dim=0).to(device=model_management.intermediate_device(), dtype=model_management.intermediate_dtype())
                 extra[k] = v
 
             r = r + (extra,)

comfy_api_nodes/apis/gemini.py

@@ -67,6 +67,7 @@ class GeminiPart(BaseModel):
     inlineData: GeminiInlineData | None = Field(None)
     fileData: GeminiFileData | None = Field(None)
     text: str | None = Field(None)
+    thought: bool | None = Field(None)
 
 
 class GeminiTextPart(BaseModel):

comfy_api_nodes/nodes_gemini.py

@@ -63,7 +63,7 @@
       $m := widgets.model;
       $r := widgets.resolution;
       $isFlash := $contains($m, "nano banana 2");
-      $flashPrices := {"1k": 0.0696, "2k": 0.0696, "4k": 0.123};
+      $flashPrices := {"1k": 0.0696, "2k": 0.1014, "4k": 0.154};
       $proPrices := {"1k": 0.134, "2k": 0.134, "4k": 0.24};
       $prices := $isFlash ? $flashPrices : $proPrices;
       {"type":"usd","usd": $lookup($prices, $r), "format":{"suffix":"/Image","approximate":true}}
@@ -188,10 +188,12 @@ def get_text_from_response(response: GeminiGenerateContentResponse) -> str:
     return "\n".join([part.text for part in parts])
 
 
-async def get_image_from_response(response: GeminiGenerateContentResponse) -> Input.Image:
+async def get_image_from_response(response: GeminiGenerateContentResponse, thought: bool = False) -> Input.Image:
     image_tensors: list[Input.Image] = []
     parts = get_parts_by_type(response, "image/*")
     for part in parts:
+        if (part.thought is True) != thought:
+            continue
         if part.inlineData:
             image_data = base64.b64decode(part.inlineData.data)
             returned_image = bytesio_to_image_tensor(BytesIO(image_data))
@@ -931,6 +933,11 @@ def define_schema(cls):
             outputs=[
                 IO.Image.Output(),
                 IO.String.Output(),
+                IO.Image.Output(
+                    display_name="thought_image",
+                    tooltip="First image from the model's thinking process. "
+                    "Only available with thinking_level HIGH and IMAGE+TEXT modality.",
+                ),
             ],
             hidden=[
                 IO.Hidden.auth_token_comfy_org,
@@ -992,7 +999,11 @@ async def execute(
             response_model=GeminiGenerateContentResponse,
             price_extractor=calculate_tokens_price,
         )
-        return IO.NodeOutput(await get_image_from_response(response), get_text_from_response(response))
+        return IO.NodeOutput(
+            await get_image_from_response(response),
+            get_text_from_response(response),
+            await get_image_from_response(response, thought=True),
+        )
 
 
 class GeminiExtension(ComfyExtension):

requirements.txt

@@ -1,4 +1,4 @@
-comfyui-frontend-package==1.41.20
+comfyui-frontend-package==1.41.21
 comfyui-workflow-templates==0.9.26
 comfyui-embedded-docs==0.4.3
 torch