Key facts
| Vision input | Image understanding runs on the OpenAI-compatible chat completions endpoint with supported multimodal models |
| Example model | plugsky-gemma-4 handles text plus image inputs such as screenshots, forms and product photos |
| Model access | 30+ models from efficient to frontier tiers behind one key |
| Dedicated media endpoints | Image generation and editing endpoints are labelled coming soon |
| Deployment | Hosted, VPC, on-prem and air-gapped options on Enterprise |
| Data handling | Data encrypted at rest and in transit; API data is not used to train models |
| Audit | Per-request logs with model, tokens, latency and user; SIEM export; retention up to 7 years |
| Free plan | plugsky-micro and plugsky-lite, no card; 14-day full-access trial |
TL;DR
- Extract text and fields from scanned filings and exhibits with a citable source.
- Keep matter collections and scoped keys exactly as you do for text.
- Audit logs tie every image request to a user, model and time.
- BYOK and private deployment cover privileged material.
- Delete source images and derived text together when a matter closes.
How it works, step by step
- Choose a closed matter with scanned documents and known ground truth.
- Define the fields worth extracting — parties, dates, amounts, exhibit references.
- Run extraction with a JSON schema and keep the source image linked to each result.
- Measure accuracy and let reviewers flag illegible or ambiguous items.
- Apply the same collection and key model as text: one matter, one key.
- Configure audit export and deletion workflows for closed matters.
- Extend to live matters only after accuracy and workflow fit are proven.
Original data
Try it yourself
Open the model availability tracker →
Why legal teams need image understanding
Legal review is full of images: scanned agreements, exhibit photographs, handwritten annotations, whiteboard captures from negotiations. Someone has to read them into the record. Vision models can extract text and fields from those images, flag what is illegible and let reviewers work from a structured summary with the source one click away.
The confidentiality model does not change because the input is an image. Exhibits belong to a matter, matter data belongs in its own collection, and only the staffed team holds the key. Extraction is a review aid, not legal advice, and every output should be checkable against the image it came from.
How vision works on the Plugsky API
The vision path is a chat completion with an image in the message content. Send the image alongside the prompt using the OpenAI-compatible request shape you already use for text, and the model returns a normal completion that you can stream, constrain with JSON mode or feed into function calling. plugsky-gemma-4 is the workhorse for high-volume visual triage — screenshots, forms, product photos and document images — while frontier models handle harder reasoning over the same input. For legal teams, each extraction stays linked to its source image, so reviewers can verify rather than trust.
The controls legal teams should configure first
Keep images inside the same access architecture as text: a collection per matter, scoped keys for the team, and audit logs that record user, model, region and time for every request, exportable to your SIEM with retention up to 7 years. Where privilege demands it, run in your VPC, on-prem or air-gapped with BYOK through AWS KMS, Azure Key Vault, HashiCorp Vault or an on-prem HSM.
For review quality, constrain extraction with JSON mode — parties, dates, obligations, exhibit numbers — and keep the original image attached to each result so a reviewer can verify. Define deletion workflows for closed matters so both source images and derived text follow the engagement's retention policy. Image generation endpoints are labelled coming soon and are not part of this design.
A practical pilot path
Pilot on a closed matter with scanned exhibits and a known ground truth. Measure extraction accuracy on names, dates and amounts, and let reviewers rate whether the structured output saved time. Prototype on the free plan, then move to the deployment your engagement terms require. See the live pricing page for current plans.
Honest comparison
| Capability | Plugsky | Vision-only API vendor | Building in-house |
|---|---|---|---|
| Image understanding | Live on supported multimodal chat models | Usually the core product | You host and serve vision models |
| Image generation | Labelled coming soon | Often available today | Custom deployment work |
| API shape | OpenAI-compatible chat completions | Vendor-specific SDK | You standardise it yourself |
| Model choice | 30+ models behind one key | Narrower catalogue | One model per integration |
| Deployment | Hosted, VPC, on-prem and air-gapped | Usually hosted only | Your infrastructure only |
| Controls | Region pinning, BYOK, per-request audit logs, SIEM export | Varies by vendor | You build the pipeline |
Frequently asked questions
Can extraction replace document review?
No. It accelerates reading and structuring; qualified reviewers still verify against the source. Output is an aid, not legal advice.
How are scanned exhibits kept privileged?
They stay in a matter-scoped collection with keys limited to the staffed team, in a deployment that meets the engagement terms, with audit logs recording every access.
Do we have to change our application code?
No. Image understanding uses the same OpenAI-compatible chat completions endpoint, so your SDK, auth and error handling carry over; the message content simply includes an image.
Is there a free plan?
Yes — plugsky-micro and plugsky-lite with 2 API keys and no credit card. A 14-day full-access trial is also available.
How is pricing structured?
Self-serve plans are flat monthly with unlimited fair-use usage and no per-token charges or overage fees. See the live pricing page for current plans.
Is image generation available?
Not yet — image generation and editing endpoints are labelled coming soon. Image understanding is live today on supported multimodal chat models.