How MNKI Works: Modules, Engines, Generation Strategies and Credits Explained
Quick answer: MNKI is a browser-based AI studio for architectural and interior visualization. You upload an input (a sketch, floor plan, massing model, wireframe, white model, photo or other drawing), pick one of nine modules, choose an image engine and a generation strategy, and generate. You can start free with 40 trial credits, and each render costs 4 to 8 credits depending on the engine and resolution. This page explains each part in plain terms so you know what you are choosing and why.
Written by the MNKI team from MNKI's own documentation and pricing page. Last verified: September 29, 2026. Because MNKI publishes this page, it describes MNKI's own product and is not an independent review; for comparisons with other tools see our comparison of AI architectural rendering tools.
The workflow at a glance
- Choose a workflow. Studio is the professional workspace (projects, folders, the modules below, camera control, history and version tracking, exports). AChAt is a conversational mode where you describe what you want in plain language and refine through dialogue. You can switch later without losing Studio projects.
- Create a project and upload an input image.
- Pick a module and an engine.
- Set style, lighting, preset, camera and strategy (or accept the defaults).
- Generate, review the result in the history bar, refine one setting at a time, and download.
Inputs MNKI accepts
MNKI lists these input types: floor plans, hand sketches, 3D massing, 3D wireframes, white models, raw photos, elevation drawings, section drawings, blueprints, site context images, perspective views, detail drawings and screenshots or exports from BIM models. Uploads can be PNG, JPEG or WebP. You can also draw directly in MNKICanvas (beta), a built-in drawing tool, and send the result straight to the Studio.
The nine modules
| Module | What it does |
|---|---|
| General Render | The all-purpose module: applies style, lighting and materials to any input |
| Architecture Study | Emphasises design intent (proportion, massing clarity, material reading) over photorealistic finish |
| Mood Board | Generates a collage-style board of textures, materials and tones |
| Material Palette | Generates a structured study of material samples, finishes and colour relationships |
| Space Planning | Generates a furnished, labelled plan from a floor plan (requires a descriptive prompt) |
| 3D Plan | Turns a 2D plan into a furnished axonometric or isometric 3D plan view |
| Inpaint Remove | Masks an element and removes it, filling in the background |
| Inpaint Add | Masks a region and adds an element you describe |
| Inpaint Replace | Masks an element and replaces it with something else |
Practical guides for the main ones: sketches to renders, 2D floor plans, interior rooms, mood boards and palettes and inpainting.
Modeling and drafting apps: 3D Sandbox and MNKICanvas
Beyond AI rendering, MNKI includes two apps that are listed on its Free plan:
- MNKI 3D Sandbox is a 3D builder (built on the open-source Pascal editor, per MNKI): draw walls, place doors and furniture, and walk through your design. Its Present tools include a sun study and view capture, and it exports DXF and PDF drawings from the toolbar (you choose which levels and section cuts to include). Its interface is available in English only.
- MNKICanvas (Beta) is a built-in drawing tool with brush, pen, eraser, line and shape tools, text and crop, a layer system, up to 50 undo steps, grid and snap-to-grid, and export to PNG (with a transparent-background option) or JPEG. You can send a drawing straight to the Studio or to AChAt for rendering.
Two image engines
| Engine | Positioning (from MNKI's FAQ) | Credits per render |
|---|---|---|
| MNG 3 Ultra | The best-quality engine | 6 credits at 2K, 8 credits at 4K |
| MNF 2.5 | A faster engine with good quality | 4 credits |
MNKI describes MNG as powered by Google Gemini and MNF as powered by Black Forest Labs' Flux models. One pattern to consider is to explore with the faster engine and finalise one or two directions on the higher-quality engine. If you are unsure, MNKI's own advice is to choose MNG when quality is the priority and MNF when speed matters.
Generation strategies
MNKI's tutorial describes the generation strategy as a dial for how closely MNKI follows your input and settings versus adding its own interpretation. There are three:
- Strict: deterministic. MNKI closely follows your input geometry, module settings and prompt with minimal interpretation. MNKI lists technical documentation, permit renders and batch consistency across a project series as best uses, and says it is not ideal for rough or incomplete inputs.
- Balanced: the default and recommended setting for most projects. It follows your input faithfully while allowing enhancements such as added context or enriched materials.
- Creative: uses your input as inspiration and adds interpretation beyond the brief. MNKI expects unpredictable results and suggests running several renders and choosing the best.
Style, lighting, presets and camera
- Styles: 28 named render styles (plus a "None" option) and 8 one-click presets, which together make up MNKI's "30+ style options". Named styles range from Modern and Scandinavian to Brutalist, Japandi, Art Deco, Coastal and Parametric.
- Lighting: options include Golden Hour, Direct Sunlight, Overcast/Diffuse, Night, Moody/Cinematic and Morning Mist.
- Presets: Interior, Exterior, Technical, Organic, Sketch, Landscape, MasterPlan and Render ID apply a set of style, lighting and material-detail settings with one click.
- Camera control (designed for exterior views): azimuth, elevation, distance, focal length from 16 mm to 135 mm, and composition rules, with presets such as Front Elevation, Hero Corner, Aerial, Isometric, Worm's Eye and Drone View.
- Reference image: upload one to match a look, palette and mood. When it is active, the manual style and lighting selectors are disabled to avoid conflicts.
- 360 panorama: an experimental feature for VR-style panoramic images that MNKI says works best with interior spaces.
Credits and plans (as listed on MNKI's pricing page)
| Plan | Price | Credits | Notes |
|---|---|---|---|
| Free | 0 | 40 trial credits, then pay-as-you-go top-ups | Top-up: 150 credits for 10.00 USD; output up to 4K per the pricing page; projects retained for 90 days |
| Basic | 28.99 USD/month | 500 credits per month | Up to 2K output; projects retained for 365 days |
| Pro | 148.99 USD/month | 3,000 credits per month | Up to 4K, batch processing; projects retained while the subscription is active |
| Team & Enterprise | Contact sales | Custom | Custom credits, API access, volume discounts |
MNKI states that all plans include unlimited projects and no watermarks. The pricing page also lists MNKIspace team workspaces (Free plans can join as a member; Basic includes one owned workspace and Pro up to five) and MNKIpresent AI presentations on Pro. MNKI's FAQ says MNKI Video (beta) is available on Pro and VIP plans. Prices may be displayed in your local currency where available (per MNKI's FAQ), and payments are processed by Noon Payments. Always confirm current details on the pricing page, as plans change.
What happens when a generation fails
Per MNKI's FAQ, you are not charged for failed generations caused by server-side issues; any deducted credits are automatically reimbursed to your MNKI wallet within one to three working days. If you get a severely distorted result, MNKI asks you to email support@mnki.ai with a screenshot so the team can review it for refund eligibility. In other words, refunds for poor-quality outputs are reviewed, not automatic. Per MNKI's tutorial, if an AI provider's safety filter refuses a prompt the generation is still billed, because the model ran, so rephrase rather than retry the same wording.
Limits worth stating up front
- MNKI's FAQ notes AI can occasionally be unpredictable. MNKI describes Strict as deterministic, but always check results against your source.
- Results are illustrations of design intent, not dimensioned drawings, engineering documents or code-compliance checks.
- Detailed proportions, materials and small features should always be checked against your source drawings.
- Consider telling clients when imagery is AI-generated. The AIA's AI Firm Toolkit recommends disclosure where AI played a substantive role in a deliverable the client relies on.
FAQ
What is MNKI?
MNKI is a browser-based AI studio that transforms sketches, floor plans, 3D models, wireframes and photos into photorealistic architectural and interior visualizations. It also includes inpainting, mood boards, material palettes, presentations, team workspaces and a beta video tool.
What is the difference between the MNG and MNF engines?
MNG 3 Ultra is MNKI's best-quality engine (6 credits at 2K, 8 credits at 4K). MNF 2.5 is faster and costs 4 credits per render.
Which generation strategy should I use?
Per MNKI's tutorial, start with Balanced. Switch to Strict if the render adds details you did not request, ignores specific elements of your floor plan or gives inconsistent results across a batch. Switch to Creative if Balanced results feel too safe and you want to explore.
How much does a render cost in credits?
4 credits for a standard render on MNF, 6 credits for a 2K render on MNG and 8 credits for a 4K render on MNG, per MNKI's pricing page. Inpaint operations cost 8 credits each, per MNKI's FAQ. Check the pages for the current table.
Can I try MNKI for free?
Yes. The free plan includes 40 trial credits and no watermark, and you can buy top-up credits without subscribing.
Sources
- MNKI pricing: plans, credits, credit costs, retention (verified September 29, 2026).
- MNKI FAQ: modules, engines, strategies, camera, failed-generation and refund policy (verified September 29, 2026).
- MNKI tutorials: input types, module descriptions, strategies (verified September 29, 2026).
- Product names and trademarks belong to their owners and are used for identification only.