Concurrent Submit | Zhenzhen_Flux_2_Pro

Between cheap Flex and premium Max sits Pro - the tier most people land on for serious-but-frequent image work: better than Flex on fidelity, less punishing than Max on price, and still able to chew through multi-image edits that a local card would choke on. This node is that endpoint resold by the Comfyui-zhenzhen pack, wrapped so you can fire a whole batch of Pro jobs in parallel instead of standing in line.

Like every node in this pack, it's an API wrapper: no weights, no VRAM, no downloads. You hand it a prompt (and up to eight reference images), it calls BFL's API through the author's shop, polls until the job lands, and hands you back a normal ComfyUI IMAGE. The KB's external-API-nodes essay calls this the "closed model through a reseller" pattern, and that's exactly the shape: paid token, metered per job, nothing local.

How the concurrent bit works

The regular Comfly_Flux_2_Pro node blocks until BFL answers. This Submit variant runs it in a shared background pool - 30 workers for image jobs across the pack - and returns a task immediately. You collect later via Concurrent Collect Images (30): feed it task_1, task_2, … and it waits for the whole batch, then outputs the images in slot order plus a status string. failure_mode on the collector picks fail-fast (default) or placeholder-on-error.

The upshot: render a dozen Pro images in one pass, not one-at-a-time. For paid API work the difference between "batch" and "series" is the difference between one coffee break and four.

The inputs that matter

Like Max, Pro keeps the surface small - BFL handles guidance and steps internally.

• prompt - the one required input.
• input_image … input_image_8 - up to eight references for editing and character consistency.
• width / height - up to 6000px in steps of 8. Remember resolution drives cost.
• safety_tolerance - 0–5, higher is more permissive.
• output_format - png or jpeg.

Output is a single task (COMFLY_IMAGE_FUTURE) that wires into the Collector.

Installing it

Part of 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 models to fetch. You need a paid token from the author's storefront (ai.t8star.org), pasted into api_key. Blank key = auth errors, no image.

Gotchas

The README's standard warnings apply: overseas endpoint means mainland users need a proxy (443 errors, the author suggests v2rayN with Tun mode), and 500s on API calls are often upstream BFL load - retry before debugging your graph. Cost-wise, Pro is metered per job and per resolution, so the shop's current pricing is the real spec sheet. And note the seed here is a ComfyUI cache seed, not something BFL's API necessarily honors - set it Fixed and you'll get the cached result rather than a fresh draw.