Catch damaged pieces with labels your inspectors already use
Superlinked gives your agent one API to score product photos against your own grade labels, so damaged pieces go to review before they ship.
- whole undamaged gum pellet
- broken or damaged gum pellet
Passed VisA: normal
- whole undamaged gum pellet
- broken or damaged gum pellet
Flagged for review VisA: chunk of gum missing
No broken piece got through, and no whole piece was flagged
VisA: normal
- whole undamaged cashew
- broken or damaged cashew
Passed
VisA: corner or edge breakage
- whole undamaged cashew
- broken or damaged cashew
Flagged
Fryum wheel
VisA, Amazon Science, CC BY 4.0
VisA: normal
- whole undamaged fryum wheel
- broken or damaged fryum wheel
Tie
VisA: corner or edge breakage
- whole undamaged fryum wheel
- broken or damaged fryum wheel
Flagged
Pipe fryum
VisA, Amazon Science, CC BY 4.0
VisA: normal
- whole undamaged snack tube
- broken or damaged snack tube
Passed
VisA: corner and edge breakage
- whole undamaged snack tube
- broken or damaged snack tube
Flagged
Chewing gum
VisA, Amazon Science, CC BY 4.0 VisA: normal
- whole undamaged gum pellet
- broken or damaged gum pellet
Passed
VisA: chunk of gum missing
- whole undamaged gum pellet
- broken or damaged gum pellet
Flagged
The board shows one pair per product. Across all 16 recorded photos the reranker flagged 8 of 8 damaged pieces and passed 7 of 8 whole ones, scoring 0.562 on both labels for the fryum wheel it tied on. SigLIP 2 base, the cheaper model, sorted 12 of 16: it passed 2 damaged pieces and flagged 2 whole ones.
Grade your own photos with your own labels, no training set
from pathlib import Pathfrom sie_sdk import SIEClientclient = SIEClient( api_key="API keysk-sie-…", base_url="https://api.superlinked.com",)labels = [ "labelwhole undamaged cashew", "labelbroken or damaged cashew",]image = {"data": Path(imagecashew-damaged-014.jpgbrowse).read_bytes(), "format": "jpeg"}items = [{"id": str(i), "text": label} for i, label in enumerate(labels)]ranked = client.score("modelQwen/Qwen3-VL-Reranker-2B", {"images": [image]}, items)print([(labels[int(row["item_id"])], row["score"]) for row in ranked["scores"]])import { readFile } from 'node:fs/promises';
import { SIEClient } from '@superlinked/sie-sdk';
const client = new SIEClient('https://api.superlinked.com', {
apiKey: 'sk-sie-…',
});
const labels = [
"whole undamaged cashew",
"broken or damaged cashew",
];
const image = await readFile("cashew-damaged-014.jpg");
const items = labels.map((text, i) => ({ id: String(i), text }));
const ranked = await client.score('Qwen/Qwen3-VL-Reranker-2B', { images: [image] }, items);
console.log(ranked.scores.map(({ itemId, score }) => ({ label: labels[Number(itemId)], score })));image_bytes=$(base64 < 'cashew-damaged-014.jpg' | tr -d '\n')
curl https://api.superlinked.com/v1/score/Qwen%2FQwen3-VL-Reranker-2B \
-H "Authorization: Bearer sk-sie-…" \
-H "Content-Type: application/json" \
-d "{\"query\":{\"images\":[{\"data\":\"$image_bytes\",\"format\":\"jpeg\"}]},\"items\":[{\"id\":\"0\",\"text\":\"whole undamaged cashew\"},{\"id\":\"1\",\"text\":\"broken or damaged cashew\"}]}"Build the "Image classify" capability into my app using the Superlinked Inference Engine (SIE).
Context
- SIE is an OpenAI-style inference API. Python SDK: `from sie_sdk import SIEClient`; TypeScript: `@superlinked/sie-sdk`.
- Base URL: https://api.superlinked.com (or my regional endpoint). Auth: Bearer key from env `SIE_API_KEY` (never hard-code it).
- Model: Qwen/Qwen3-VL-Reranker-2B (SIE primitive: /score). Keep the model id configurable.
Task
- Input: an uploaded image plus candidate labels.
- Behaviour: return the candidate image labels ranked by similarity
- Score the image against the candidate-label items, then rank labels by score.
Deliverables
- A typed client wrapper, an application-level function for this task, error handling for timeouts/empty input, and unit tests with a stubbed client.
- Wire it into my existing stack (ask me which framework if unclear) and add a short usage example. Edit the code and run it live Sign up to join the waitlist. We'll notify you when access is available. Credits are handled separately.
Get started no credit card required
Output
broken or damaged cashew0.637
whole undamaged cashew0.547
Top label broken or damaged cashew at 0.637 across 2 candidates
- No image-specific training data needed
Image classification quality and latency
PRICE
$ / 1k images $ / 1M input tokens $ / 1M pairs ↓
$0.0143 SIE SigLIP2 Base
$0.149 SIE Qwen3 VL Reranker 2B
AWS Rekognition $1
Google Vision Labels $1.5
Google Vision Objects $2.25
$0.023 SIE SigLIP2 Base
OpenAI GPT-5.4 mini $0.75
SIE Qwen3 VL Reranker 2B $222.79
QUALITY
Accuracy ↑
SIE Qwen3 VL Reranker 2B 0.91
0.89
0.88
SIE SigLIP2 Base 0.86
0.83
0.82
LATENCY
p50 ms ↓
35ms SIE Qwen3 VL Reranker 2B
45ms
55ms
SIE SigLIP2 Base 99ms
120ms
135ms
Deploy your way
Managed Cloud
Full compute toolkit for your agents with zero ops.
- No idle GPUs, pay for what you use
- Fits your stack: SDK, API, CLI, MCP
- Zero lock-in, self-host the same stack
- SOC 2 Type 2, US or EU data residency
$50 Example credit balance
1.3B tokens embedded
120k pages parsed
27M tokens generated (1:1 in:out)
Get started
no credit card required
Self-host with K8s
Easy & scalable deployment in your own cloud.
- Terraform to your cloud in minutes
- Apache-2.0, same engine as Cloud
- Scales to zero, no bill between jobs
- Per-tenant pools, no noisy neighbors
Agent prompt
Deploy SIE to our AWS account with the superlinked/sie/aws Terraform module. Docs: superlinked.com/docs/deploymentDeploy SIE to our GCP project with the superlinked/sie/google Terraform module. Docs: superlinked.com/docs/deploymentDeploy SIE to our Azure AKS cluster via helm install. Requirements: superlinked.com/docs/deployment Run locally
Run the same models on your own machine.
- Runs on NVIDIA GPU or Apple Silicon
- One command, no Docker or cluster
- All 100+ Cloud models, fully offline
- Same SDK and IDs, no code changes
pip install "sie-server[local]" && sie-server servepip install "sie-server[local]" && sie-server serve --device cuda