Create posters, ecommerce visuals, designs and comics through API or MCP

AI image generation, editing and scale

• Generate and edit

  Start from a prompt or reference for posters, ecommerce, design and comic images, then change backgrounds, recolor or derive variations.

• Concurrent workloads

  Website, API and MCP use separate concurrency pools by default. The current limit comes from account entitlements, and administrators can switch an individual user to one shared pool.

• Affordable image generation

  Enabled image channels such as GPT Image 2, Nano Banana, Gemini 3.8 Flash, 3.5 Flash-Lite and 3.1 Pro come from the live catalog, which shows credits and RMB or USD estimates and confirms the maximum charge first.

• Multiple agent platforms

  Remote MCP supports Codex, WorkBuddy, QwenWork and Doubao Work for image generation and editing in familiar agent workspaces.

Official model names and InkRoom API model IDs are shown separately

Callable entries below come only from the current environment's live image-model catalog. GPT Image 2, Nano Banana, Gemini 3.8 Flash, 3.5 Flash-Lite, 3.1 Pro and other channels appear only when enabled and marked image-capable. Channel API model IDs remain visible and are not promoted as official Google image-model names.

How does Google currently name its image models?

Google's image-generation products include Nano Banana 2 (Gemini 3.1 Flash Image), Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) and Nano Banana Pro (Gemini 3 Pro Image). gemini-3.8-flash is a callable InkRoom channel ID. InkRoom labels callable channel IDs separately and uses the live catalog to identify image-capable channels.

• Read Google's image-generation documentation

  Read Google's image-generation documentation

• Read the OpenAI GPT Image 2 documentation

  Read the OpenAI GPT Image 2 documentation

Two access methods with the same real models and billing

The public APIs and MCP are separate protocols, but both use InkRoom's live model catalog, account entitlements and credit billing.

• Model APIs

  Use the OpenAI-compatible /v1 or Gemini-compatible /v1beta API with standard SDKs for batch or concurrent generation and reference-image edits.

  View API docs

• InkRoom Remote MCP

  Connect Codex, WorkBuddy, QwenWork or Doubao Work to generate, edit and create poster or ecommerce images with cost preview and confirmation retained.

  Set up MCP

How are low entry prices and concurrency confirmed?

Pricing is not a hard-coded marketing number. The model directory shows credits and RMB or USD estimates for each enabled model, and confirms the cost again from model, resolution, count and other settings before generation. Website, API and MCP have separate concurrency pools by default; an administrator can configure one shared pool for an individual user. Check your current limit and usage before submitting.

Try image generation for free

When a guest trial is available, the page shows remaining uses and models. Only a successful image consumes one use.

Frequently asked questions

Which protocols does the InkRoom image API support?

InkRoom provides an OpenAI-compatible /v1 API and a Gemini-compatible /v1beta API. The public API does not require an extra mandatory InkRoom-private header.

Are API and MCP concurrency limits separate?

Yes, by default. Website, API and MCP each use the account concurrency limit; administrators can switch an individual user to one shared limit.

Can an MCP token be used as an API key?

No. MCP tokens and public API keys are separate credentials for Remote MCP and the /v1 or /v1beta public APIs respectively.

Can Codex, WorkBuddy, QwenWork and Doubao Work generate and edit images through MCP?

Yes. All four connect to the same InkRoom Remote MCP for prompt-based generation, reference-image edits, posters and ecommerce visuals with a cost preview before execution. Codex and WorkBuddy can use agent-managed setup; QwenWork and Doubao Work use their connector interfaces.

Is the InkRoom API suitable for high-volume image generation?

Yes, for batch and concurrent image workloads. Actual concurrency comes from account entitlements and backend configuration; the live catalog shows credits and RMB or USD estimates, with the maximum charge confirmed before submission.

How do Google's official image-model names differ from InkRoom API model IDs?

Google currently uses names such as Nano Banana 2 (Gemini 3.1 Flash Image), Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) and Nano Banana Pro (Gemini 3 Pro Image). InkRoom separately shows callable channel model IDs in the live catalog; it does not force an equivalence, and the catalog determines actual availability.

Can I use affordable GPT Image, Nano Banana, Gemini 3.8, 3.5 or 3.1 Pro for images in InkRoom?

You can use image channels enabled in the current environment. GPT Image 2 and Google's Nano Banana family are official image-model names; Gemini 3.8 Flash, 3.5 Flash-Lite and 3.1 Pro are shown as live API model IDs and are not automatically presented as official Google image models. A model can generate images only when the live catalog marks it as image-capable; capabilities and prices come from that catalog.

Can I try image generation for free first?

Eligible guests can try text-to-image for free. Remaining uses and models are shown before prompt entry; availability, daily limits and models follow live configuration, and only successful images use the allowance. Reference-image editing requires sign-in and follows account access and live pricing.