Concurrent Submit | Zhenzhen GPT Image 2 Fal

GPT Image 2 is OpenAI's current image model, and this node is the pack's fal-hosted channel for it - the one the README bills as fast, stable, and 4K-capable with unlimited concurrency. The -fal suffix means the job runs through fal's queue API on the author's reseller backend, and per the pack's notes it's pre-charge: credits are withheld on submit and settled when the job finishes. Where the official-channel GPT Image 2 nodes can be slow and rate-limited, this channel was added specifically as the fast backup - the README claims 4K in about 70 seconds.

The Concurrent Submit wrapper is a natural fit: "unlimited concurrency" is only useful if you can actually fire a pile of jobs at once, and that's exactly what this node does.

How the concurrent bit works

The plain Comfly_gpt_image_2_fal node blocks until fal returns. The Submit variant runs it in the pack's shared image pool (30 workers) and returns a task immediately. Collect via Concurrent Collect Images (30): wire task_1 … and the collector waits for the batch, returning images in slot order plus a status string. failure_mode = fail-fast (default) or placeholder.

The inputs that matter

• prompt (required) - the description or edit instruction.
• mode - generate or edit. In edit mode, connect images (and optionally a mask).
• image1 … image4 - reference images for edits.
• mask - optional inpainting mask.
• quality - high/medium/low/auto.
• image_size - auto, the standard presets, or custom. Custom unlocks custom_width / custom_height up to 3840 - must be multiples of 16, or the call errors. Defaults are 3840×2160, i.e. a 4K canvas.
• num_images - 1–4 per call.
• image_way - image_url (default) or base64 for how references are uploaded.
• seed - fal caps at 65535; 0 = random.
• poll_interval (6s) / max_poll_attempts (600 → 3600s ceiling) - fal polling knobs.

Output: one task (COMFLY_IMAGE_FUTURE) → Collector → images.

Installing it

Ships in Comfyui-zhenzhen:

• ComfyUI Manager → Install Custom Nodes → search "Comfyui-zhenzhen" → install → restart.
• Or cd ComfyUI/custom_nodes && git clone https://github.com/T8mars/Comfyui-zhenzhen and restart.

No model downloads. You need a paid token from the author's storefront (ai.t8star.org), in api_key.

Gotchas

The README is explicit on three points. First, fal is pre-withheld: don't be alarmed when a job's hold is bigger than the final charge. Second, the custom-size rule - must divide by 16 - is the #1 cause of "it just errored" for people chasing specific resolutions. Third, this fal channel was the fix for the official GPT Image 2 channel's congestion, so if the official nodes are slow, this is the one to reach for. The seed cap (65535, 0 = random) also trips people up when porting seeds from other nodes. Standard pack note: overseas endpoint needs a proxy from mainland China.