ViewComfy compute and growth docket#

Status: Decision log Scope: Managed ComfyUI deployment and workflow-to-app tooling, not the official Comfy product.

2026-10-04 — Workflow packaging and collaborative distribution#

Compute and procurement#

Published offer: GPU time is billed by active second, with CPU and RAM separately priced. The current table lists T4 $0.65/hour, L4 $1.16, A10G $2.20, L40S $3.51, A100 40 GB $4.10, A100 80 GB $6.15, H100 $8.42, H200 $9.65 and B200 $13.32. These are customer rates, not supplier costs; no RTX PRO 6000 rate is listed. The table alone does not disclose hardware ownership or GPU supplier. Pricing.

Engineering disclosure, 2025-07-22: CPU/RAM snapshots shorten Comfy startup; the company's example workflows improve from 53–77 seconds to 8–10 seconds. Model loading is explicitly excluded, nodes using GPUs during initialization are incompatible, and enablement required contacting the team at publication. These are vendor examples, not our benchmark or proof of today's default. Snapshot explanation.

Enterprise deployment can use hosted infrastructure or customer on-premises/offline environments. A customer-owned deployment option is not proof ViewComfy owns its hosted fleet. Enterprise.

Acquisition and monetization#

Current offer: Developer has a $0 platform fee plus compute; Team is $30/month plus compute with $10 monthly compute credits; Studio is $200 plus compute with $30 credits. App counts, collaboration and concurrency differentiate tiers; additional developer seats are separately priced. A free platform tier is not free inference. Plan boundaries.

The open-source app builder turns workflows into interfaces accessible to nontechnical colleagues, with app sharing and branding. Inference: embedding repeatable workflows into a team's work may drive distribution and paid collaboration, rather than relying on unlimited free image generation. Neither the page nor captured pricing establishes a conversion cohort. Workflow-to-webapp product.

Outcomes, implication and unknowns#

No audited active-user, revenue or fleet-utilization series was established in this pass. The useful comparison is deployment reproducibility, startup/load latency and customer app delivery. Purchased versus leased GPUs, actual hosted GPU provider, negotiated costs, cancellation behavior and retained paid accounts remain unknown; no allocation or checkout was tested.

Sources retrieved October 4 local; undated offers are snapshots. Batch manifest, coverage index.