Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
24 commits
Select commit Hold shift + click to select a range
011783a
[Partner Nodes] feat(Pika): add Pika H3 Flex Video and Reference to V…
bigcat88 Nov 4, 2026
685c49b
[Partner Nodes] feat(Pika): add Pika Soundtrack, SFX and Music nodes …
bigcat88 Nov 4, 2026
d4035aa
[Partner Nodes] feat(Tencent): add Tencent HY LLM node (#17534)
bigcat88 Nov 4, 2026
efb201b
[Partner Nodes] feat(Kling): add Kling 4.0 Flash model and original a…
bigcat88 Nov 6, 2026
5a6d2b5
[Partner Nodes] feat(OpenAI): add GPT Image 2.6 Flare and Sunburst mo…
bigcat88 Nov 6, 2026
4f90883
[Partner Nodes] feat(Qwen): move Qwen Image nodes to Comfy Router (#1…
bigcat88 Nov 6, 2026
0d44271
[Partner Nodes] feat(Gemini): add Omni Flash 1.2 to the Omni video no…
bigcat88 Nov 6, 2026
c28e2c6
Update workflow templates to v0.11.103 (#17563)
comfyui-wiki Nov 6, 2026
49b61c3
ComfyUI v0.44.1
Nov 6, 2026
2f796b5
Bump comfyui-frontend-package to 1.58.5 (#17442)
comfy-pr-bot Nov 4, 2026
3f2ebd1
Add better error for when pixel_space VAE wrongly selected. (#17518)
comfyanonymous Nov 4, 2026
a0d4d66
Use adaptive slicing in seedvr2 decode (#17508)
kijai Nov 4, 2026
6000263
Turn on local_web_sso and partner_node_token by default (#17540)
dante01yoon Nov 5, 2026
23d79a3
Update comfy-kitchen version to 0.2.41 (#17542)
comfyanonymous Nov 5, 2026
f83b20b
Fix potential crash. (#17543)
comfyanonymous Nov 5, 2026
a41b69a
Performance improvements for qwen 3.x (Part 4) - Combine Parallel Li…
rattus128 Nov 5, 2026
f620418
Name comfy-agent without a version in agent_requirements.txt (CORE-69…
christian-byrne Nov 5, 2026
fddd76e
Fix Remesh Mesh on imported meshes (#17531)
kijai Nov 5, 2026
abee7dd
Support RelayVSR (CORE-701) (#17511)
kijai Nov 5, 2026
ce2b186
Fix resolution selector megapixel precision (#17555)
GACLove Nov 6, 2026
4c2eedf
Update api-node.md (#17569)
alexisrolland Nov 6, 2026
bd06f99
Move run_cpu portable script to ./advanced (#17571)
comfyanonymous Nov 6, 2026
29d2d1c
Make the partner node team code owners of comfy_api_nodes (#17575)
comfyui-wiki Nov 6, 2026
73ca416
[Partner Nodes] feat(ByteDance): add 4K resolution to Seedance 2.5 (#…
bigcat88 Nov 6, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions .ci/windows_nvidia_base_files/advanced/run_cpu.bat
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
..\python_embeded\python.exe -s ..\ComfyUI\main.py --cpu --windows-standalone-build
pause
2 changes: 0 additions & 2 deletions .ci/windows_nvidia_base_files/run_cpu.bat

This file was deleted.

3 changes: 0 additions & 3 deletions .github/PULL_REQUEST_TEMPLATE/api-node.md
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,3 @@ If **Need pricing update**:
### QA
- [ ] **QA done**
- [ ] **QA not required**

### Comms
- [ ] Informed **Kosinkadink**
1 change: 1 addition & 0 deletions CODEOWNERS
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
* @comfyanonymous @guill @alexisrolland @rattus128 @kijai
/comfy_api_nodes/ @comfyanonymous @guill @alexisrolland @rattus128 @kijai @comfyui-wiki @purzbeats

/CODEOWNERS @comfyanonymous
/AGENTS.md @comfyanonymous
Expand Down
2 changes: 1 addition & 1 deletion agent_requirements.txt
Original file line number Diff line number Diff line change
@@ -1 +1 @@
comfy-agent==0.0.0.dev0
comfy-agent
10 changes: 6 additions & 4 deletions app/agent_host.py
Original file line number Diff line number Diff line change
Expand Up @@ -48,14 +48,16 @@ def load() -> bool:
entry_points = {ep.name: ep for ep in importlib.metadata.entry_points(group=ENTRY_POINT_GROUP)}
if PROTOCOL not in entry_points:
# Some comfy-agent is installed (other protocols present) -> "update";
# nothing there -> "install".
# nothing there -> "install". The file names the package without a
# version, so only `-U` moves an old or placeholder install.
req = os.path.join(os.path.dirname(os.path.abspath(folder_paths.__file__)), "agent_requirements.txt")
install = f'pip install -U -r "{req}"'
if entry_points:
logging.warning(f"The installed comfy-agent offers agent protocols {sorted(entry_points)}; this ComfyUI needs {PROTOCOL}. Update comfy-agent to use the agent.")
logging.warning(f"The installed comfy-agent offers agent protocols {sorted(entry_points)}; this ComfyUI needs {PROTOCOL}. Update it with `{install}` to use the agent.")
_failure = {"state": "failed", "reason": "outdated"}
emit_event("agent_error", reason="outdated")
else:
req = os.path.join(os.path.dirname(os.path.abspath(folder_paths.__file__)), "agent_requirements.txt")
logging.warning(f"`--enable-agent` was passed but the comfy-agent package is not installed, so the agent is disabled. Its requirements are in {req}.")
logging.warning(f"`--enable-agent` was passed but the comfy-agent package is not installed, so the agent is disabled. Install it with `{install}`.")
_failure = {"state": "failed", "reason": "package_missing"}
emit_event("package_missing")
_publish_flag()
Expand Down
8 changes: 8 additions & 0 deletions comfy/latent_formats.py
Original file line number Diff line number Diff line change
Expand Up @@ -880,6 +880,14 @@ def __init__(self):
0.3971, 1.0600, 0.3943, 0.5537, 0.5444, 0.4089, 0.7468, 0.7744
]).view(1, self.latent_channels, 1, 1, 1)

class RelayVSR(Wan22):
# the keyframe Wan and the Flash decoder both work on the normalized latent
def process_in(self, latent):
return latent

def process_out(self, latent):
return latent

class HunyuanImage21(LatentFormat):
latent_channels = 64
latent_dimensions = 2
Expand Down
10 changes: 10 additions & 0 deletions comfy/ldm/seedvr/constants.py
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,16 @@
SEEDVR2_VAE_CACHE_QUANT_BYTES = 64 * 1024 ** 2 # tails at or above this size are packed.
SEEDVR2_CACHE_BYTES_PER_FRAME_PIXEL = 6200 # host-held caches and parked frames, per output-frame pixel.

# Free memory above the floor goes to the caches on the device first, then to longer slices.
SEEDVR2_MAX_SLICE_LATENTS = 4
SEEDVR2_SLICE_MEM_HEADROOM = 0.8
SEEDVR2_DECODE_SLICE_BYTES_PER_FRAME_PIXEL = 5300 # per latent frame of a slice, per output-frame pixel.
SEEDVR2_ENCODE_SLICE_BYTES_PER_PIXEL = 3150 # per input-frame pixel.
SEEDVR2_DECODE_TAIL_BYTES_PER_FRAME_PIXEL = 5550 # caches on the device; the fixed part covers tiles too small to pack.
SEEDVR2_DECODE_TAIL_FIXED_BYTES = 768 * 1024 ** 2
SEEDVR2_ENCODE_TAIL_BYTES_PER_PIXEL = 2700
SEEDVR2_ENCODE_TAIL_FIXED_BYTES = 512 * 1024 ** 2

# VAE decode estimate: the free VRAM an untiled decode needs under cudaMallocAsync (ComfyUI's default),
# fitted 2-7% above the measured floor from 480p to 4K. Flat in clip length: the decoded frames go
# to sd.py's output buffer off the GPU slice by slice (1088p: 5.33 GiB allocated at 61 and 121 frames).
Expand Down
166 changes: 130 additions & 36 deletions comfy/ldm/seedvr/vae.py

Large diffs are not rendered by default.

2 changes: 1 addition & 1 deletion comfy/ldm/wan/model_animate2.py
Original file line number Diff line number Diff line change
Expand Up @@ -210,7 +210,7 @@ def prefetch(self, i, device, dtype):
with comfy.model_prefetch.pause_malloc_graph():
self._staging[buf_key] = [torch.empty(t.shape, dtype=buf_key[1], device=device) for _ in range(2)]
r = self._staging[buf_key][i % 2]
stream = comfy.model_management.get_offload_stream(device)
stream = comfy.model_management.get_data_stream(device)
cs = comfy.model_management.current_stream(device)
if stream is not None and cs is not None:
# Wait for staging allocation and the previous consumer's reads before overwriting the buffer.
Expand Down
Loading
Loading