# Primate Vision API — Primate Intelligence

> Video scene understanding API. Upload a video (or point at a URL), ask a question in plain English, get a deterministic answer with confidence and clip timestamps — no hallucinations. Zero-touch provisioning: `POST https://api.primateintelligence.ai/v1/sandbox` returns an instant free test key (no email, no card).

Start here (agents): [Quickstart for AI agents](https://www.primateintelligence.ai/docs/agents.md)

## Machine-readable surfaces

- [OpenAPI 3.1 spec](https://api.primateintelligence.ai/v1/openapi.json): the full API contract — source of truth
- [Error registry](https://api.primateintelligence.ai/v1/errors): every error code with retryable/idem flags
- [Test fixture](https://api.primateintelligence.ai/v1/test-fixture): stable video + prompt + expected answer for CI self-verification
- [Credit pricing](https://api.primateintelligence.ai/v1/credit-pricing): current pricing, public — PREVIEW PRICING: early-access rate, may change up or down with model maturity, demand, and processing costs; read the live rate from GET /v1/credit-pricing (pricing_tier: "preview"); never hardcode the rate.

## Integration contract

- Base URL: `https://api.primateintelligence.ai` · Auth: `Authorization: Bearer $PRIMATE_API_KEY`
- Env var: `PRIMATE_API_KEY` (both SDKs + MCP server read it; keys never go in URLs or tool arguments)
- Core loop: `POST /v1/videos` → `POST /v1/analyses` → `GET /v1/analyses/{id}` (or `Prefer: wait=60` to collapse the poll)
- SDKs: `npm install @primate-intelligence/sdk` · `pip install primate-intelligence`
- MCP server: `npx @primate-intelligence/mcp` (env: PRIMATE_API_KEY)

## Docs

- [Quickstart](https://www.primateintelligence.ai/docs/quickstart.md): First analysis in three curl commands — no signup required.
- [Quickstart for AI agents](https://www.primateintelligence.ai/docs/agents.md): The zero-human-intervention integration path — canonical copy served by the API deploy artifact.
- [Uploading video](https://www.primateintelligence.ai/docs/guides/uploading.md): Three ways to get video in — presigned upload, URL ingest, SDK helper — and how to choose.
- [Prompts & queries](https://www.primateintelligence.ai/docs/guides/prompts.md): Free-text prompts vs structured queries, what's answerable, and how unassessable components behave.
- [Waiting for results](https://www.primateintelligence.ai/docs/guides/waiting.md): Poll vs Prefer: wait vs webhooks — pick the right completion strategy.
- [Webhooks](https://www.primateintelligence.ai/docs/guides/webhooks.md): Standard Webhooks signing, verification, retries, redelivery, and the integration rules that matter.
- [Errors & retries](https://www.primateintelligence.ai/docs/guides/errors-retries.md): The error envelope, the closed registry, and exactly what to retry.
- [Rate limits & quotas](https://www.primateintelligence.ai/docs/guides/rate-limits.md): Limit classes, headers, 429 handling, and how quotas differ from rate limits.
- [Test mode](https://www.primateintelligence.ai/docs/guides/test-mode.md): pv_test_ keys, deterministic fixtures, and CI verification without burning quota.
- [Versioning & deprecation](https://www.primateintelligence.ai/docs/guides/versioning.md): Compatibility guarantees, deprecation process, model sunsets, and service expectations.
- [Security model](https://www.primateintelligence.ai/docs/guides/security-model.md): API keys, client tokens, scopes, data handling, and the key-compromise procedure.
- [Billing & credits](https://www.primateintelligence.ai/docs/guides/billing.md): Frame-based settlement, credit grants, auto-refill, and handling insufficient_credits in code.
- [Streaming](https://www.primateintelligence.ai/docs/guides/streaming.md): Real-time video analysis over WebRTC — create, signal, go live, get per-frame results.
- [Tutorial: build from scratch](https://www.primateintelligence.ai/docs/guides/build-from-scratch.md): Zero to a working video-question CLI in ~15 minutes — no account required to start.
- [Supported actions](https://www.primateintelligence.ai/docs/supported-actions.md): What Darwin-preview-1.3B can and cannot detect — the honest capability list, including the full action vocabulary and known gaps.
- [Error registry](https://www.primateintelligence.ai/docs/errors.md): Every error code with retry semantics — the docs_url targets.
- [Changelog](https://www.primateintelligence.ai/docs/changelog.md): Dated API changes, newest-first — canonical copy served by the API deploy artifact. Subscribe via RSS at /docs/changelog.xml.
- [API reference](https://www.primateintelligence.ai/docs/reference.md): Every endpoint and schema, generated from the OpenAPI 3.1 contract.

## Full corpus

- [llms-full.txt](https://www.primateintelligence.ai/llms-full.txt): every docs page in one file

## Product pages

- [Home](https://primateintelligence.ai/): Product overview — markdown twin at /index.md
- [Pricing](https://primateintelligence.ai/pricing): $0.0000015/frame-processed metered credits — PREVIEW PRICING: early-access rate, may change up or down with model maturity, demand, and processing costs; read the live rate from GET /v1/credit-pricing (pricing_tier: "preview"); never hardcode the rate. Markdown twin at /pricing.md
- [Performance & latency](https://primateintelligence.ai/performance): Measured production p50/p95 latency, throughput, and accuracy benchmarks — markdown twin at /performance.md
- [Use Cases](https://primateintelligence.ai/use-cases): Eight markets, one model — markdown twin at /use-cases.md
- [For AI agents](https://primateintelligence.ai/agents): Zero-touch integration for AI agents — markdown twin at /agents-page.md
- [Technology](https://primateintelligence.ai/technology): How Darwin (JEPA world model) works — markdown twin at /technology.md
- [Technology — How JEPA works](https://primateintelligence.ai/technology/how-jepa-works): Joint-embedding predictive architecture explained
- [Technology — Determinism](https://primateintelligence.ai/technology/determinism): Why same-input → same-output matters for production
- [Technology — Benchmarks](https://primateintelligence.ai/technology/benchmarks): Action detection, temporal localization, published results
- [Technology — Real-time streaming](https://primateintelligence.ai/technology/real-time): WebRTC ingestion and streaming architecture
- [Team](https://primateintelligence.ai/team): Founders — markdown twin at /team.md
- [About](https://primateintelligence.ai/about): Mission, founders, founding story — markdown twin at /about.md
- [Values](https://primateintelligence.ai/values): Seven principles — markdown twin at /values.md
- [Careers](https://primateintelligence.ai/careers): Open roles — markdown twin at /careers.md
- [FAQ](https://primateintelligence.ai/faq): Frequently asked questions — markdown twin at /faq.md

## Compare

Honest, spec-table-driven comparisons of Primate Vision against every major video understanding API.
Each comparison has a markdown twin: append .md to the compare URL (e.g. /compare/twelve-labs.md).

- [Compare hub](https://primateintelligence.ai/compare): All comparisons — markdown twin at /compare.md
- [Primate Vision vs. AWS Rekognition + Bedrock Nova (2026): The Door AWS Closed](https://primateintelligence.ai/compare/aws-rekognition): AWS closed Rekognition streaming to new customers on April 30, 2026 — its own migration advice is to snapshot frames into the image API. Here's what remains, what Bedrock Nova actually offers, and the honest head-to-head against a real-time verdict API.
- [Primate Vision vs. Azure Video Indexer (2026): An Insight Catalog vs. a Live Answer](https://primateintelligence.ai/compare/azure-video-indexer): Azure Video Indexer is a genuinely good archive-enrichment product — thirty insight types, far cheaper per hour than us. But it's a catalog, not an answer: no live-stream API, no way to ask your own question of the pixels, no verdict. Here's the honest head-to-head.
- [The Best Video Understanding APIs in 2026 (an Honest Roundup)](https://primateintelligence.ai/compare/best-video-understanding-apis-2026): Six video understanding APIs — Google Video Intelligence, Twelve Labs, AWS Rekognition, Gemini, Mixpeek, and ours — scored on the six dimensions buyers actually ask about: accuracy approach, hallucination risk, pricing, streaming, free tier, and agent access. With receipts, and with our bias disclosed in the first sentence.

## Learn

- [Blog](https://primateintelligence.ai/blog): Research notes and perspectives — RSS at /rss.xml
- Every blog post has a markdown twin: append .md to the post URL (e.g. /blog/darwin-video-jepa-benchmarks.md)
- [What Scene Understanding Looks Like in a Real Security Camera Pipeline](https://primateintelligence.ai/blog/scene-understanding-security-cameras): A concrete walkthrough of what changes — and what stays the same — when you add Primate Vision to a security camera CV pipeline.
- [Primate Vision vs YOLO v8: A Head-to-Head on Real Security Camera Footage](https://primateintelligence.ai/blog/primate-vision-vs-yolo-benchmark): We ran both models on the same real-world VIRAT dataset clips. Here's what the results actually look like.
- [The False Tradeoff in Computer Vision](https://primateintelligence.ai/blog/false-tradeoff-computer-vision): For five years, CV teams have been forced to choose between reliable (but rigid) detection and flexible (but unreliable) VLMs. That tradeoff is an artefact of architecture, not an inherent property of the problem.
- [Why 'Deterministic' Matters More Than Accuracy in Production CV](https://primateintelligence.ai/blog/deterministic-cv-matters): Production CV systems don't just need models that are accurate. They need models that are predictable. Here's why that's a harder problem.
- [Technology](https://primateintelligence.ai/technology): Model releases and capability updates
- [Darwin-preview-1.3B](https://primateintelligence.ai/technology/darwin-preview): The Darwin-preview-1.3B model — full supported-actions vocabulary (532 classes) — markdown twin at /technology/darwin-preview.md

## Legal

- [Privacy Policy](https://primateintelligence.ai/privacy)
- [Terms of Service](https://primateintelligence.ai/terms)
- [Cookie Policy](https://primateintelligence.ai/cookie-policy)