API pour développeurs

API IA spécialisées

Intelligence juridique, Speech AI, NLP et Image Processing — API spécialisées pour les industries réglementées. Endpoints compatibles OpenAI, clé API unique, tarification à l'usage.

Commencer avec des crédits gratuits Contacter l'équipe commerciale

Starter en libre-service

Développez et testez avec des crédits gratuits.

Aucune carte bancaire requise. Idéal pour les développeurs qui valident un workflow et réussissent leurs premiers appels API.

Créer une clé API

Pack Growth

Engagez un volume mensuel et regroupez vos API.

Idéal pour les SaaS, les marketplaces et les équipes support qui font passer en production leurs workflows de sécurité des contenus, de NLP ou de traitement de documents.

Demander un audit de migration

Garde-fous Enterprise

Ajoutez sécurité, SLA et accompagnement architecture.

Idéal pour les équipes en environnement réglementé, les plateformes d'agents et les acheteurs à fort volume qui exigent des workflows prêts pour les services achats.

Parler à un architecte

API spécialisées

Des API mono-usage avec leur propre page dédiée, leur benchmark et leur règle de tarification — la remise découle de la position qualité par rapport à la moyenne du marché.

Brainiall Cutout

Alternative à Azure Image Analysis 4.0

KPI0.95 mIoU · ~0.4s/image
Gratuit100 images/mois
Payant à partir de$19/mois · 1k images

Voir la page et le benchmark →

Brainiall Audio Polish

Débruitage · isolation · nettoyage · mastering

KPI0.8s pour 30s d'audio
Gratuit10 minutes/mois
Payant à partir de$29/mois · 500 min

Voir la page et le benchmark →

Brainiall Speaker ID

Moteur Brainiall Speaker ID — qui a dit quoi

KPI12% DER sur AMI
Gratuit10 minutes/mois
Payant à partir de$25/mois · 300 min

Voir la page et le benchmark →

Brainiall Document Reader

Moteur Brainiall Document Reader, prise en charge des équations

KPI~3s/page · tableaux préservés
Gratuit20 pages/mois
Payant à partir de$19/mois · 1k pages

Voir la page et le benchmark →

Brainiall Identity

Alternative à AWS Rekognition Face · conforme LGPD

KPI154ms p50 · 2-3× plus rapide que Rekognition
Gratuit100 vérifications KYC/mois
Payant à partir de$0.005/vérification

Voir la page et le benchmark →

Brainiall Content Guard

NSFW + régions · 20% moins cher que Rekognition

KPI91ms p50 · 3-4× plus rapide
Gratuit1k images/mois
Payant à partir de$0.0008/image

Voir la page et le benchmark →

Brainiall OCR

Brainiall Form Parser + moteur Brainiall Form Parser · 5-10× moins cher qu'AWS Textract

KPI327ms (Fast) / 2.4s (Pro)
Gratuit500 pages/mois
Payant à partir de$0.0015/page

Voir la page et le benchmark →

Brainiall Vision Tagger

Moteur Brainiall Vision Tagger multi-tâches · zero-shot (contrairement à Rekognition)

KPI236ms (Fast) / 1.6s (Standard)
Gratuit500 images/mois
Payant à partir de$0.0015/image

Voir la page et le benchmark →

Brainiall Memory

Alternative à Mem0, à 1/5 du coût

KPIRecall@5 95% · 31ms p50
Gratuit10k événements/mois
Payant à partir de$19/mois · 1M événements

Voir la page et le benchmark →

NLP Suite

Your team is building text analysis from scratch — or overpaying cloud NLP APIs. Ten text & language APIs at half the price of Google or AWS. Toxicity, sentiment, entities, PII, language — all under 200ms.

<200msNo Cold StartsPay Per Request

What you can do

Keep your platform safe — automated content moderationUnderstand what your customers are saying at scaleStay GDPR/CCPA compliant — automatic PII anonymizationAutomated compliance screeningIntelligent multilingual routingSummarize long documents & answer questions about themManaged RAG — ingest your docs, ask questions, get cited answersTranslate content between 100+ languages

Endpoint	Brainiall	Google NLP	AWS Comprehend	Azure Text Analytics
Toxicity	$0.001/req	$0.002/req	$0.003/req	$0.002/req
Sentiment	$0.001/req	$0.002/req	$0.003/req	$0.002/req
Entities	$0.002/req	$0.002/req	$0.003/req	$0.002/req
PII Detection	$0.002/req	N/A	$0.003/req	$0.002/req
Language	$0.0005/req	$0.002/req	$0.001/req	$0.001/req
Translation (100+ languages)	$5/1M chars	~$20/1M chars	~$15/1M chars	~$10/1M chars
Summarization (extractive + abstractive)	$0.001/1K chars	N/A	N/A	~$0.001/1K chars (AI Language)
Grounded Q&A	$0.001/1K chars	N/A	N/A	~$0.001/1K chars (AI Language)
Knowledge base — ingest (managed RAG)	$0.001/1K chars	N/A	N/A	varies (AI Search SU/hr)
Knowledge base — query (retrieval + grounded answer)	$0.005/query	N/A	N/A	varies

50% cheaper than Google NLP. No minimum spend. No contracts. First API call in 60 seconds.

Try NLP Suite — No Contracts

Speech AI

Building voice features in-house takes months and costs a fortune. Production-ready voice APIs — pronunciation scoring, speech-to-text, text-to-speech, and transcription in 99+ languages. Ship in days, not months.

<200msReal-Time Streaming99+ Languages

What you can do

Ship a language learning feature in days, not monthsTranscribe support calls automaticallyAdd voice accessibility to your productAutomate podcast/meeting transcriptionPhoneme-level pronunciation scoringVoice assistants in 99+ languagesVoice authentication & speaker identification

Endpoint	Brainiall	Google Speech	AWS Transcribe	Azure Speech
Pronunciation	$0.01-0.02/req	N/A	N/A	$0.02/req
STT	$0.01-0.02/req	$0.016/min	$0.024/min	$0.016/min
Voice (Edge tier)	$0.015/1K chars	$0.016/1K chars	$0.016/1K chars	$0.016/1K chars
Voice (Pro tier — cloning, 99 langs, 48 kHz)	$0.054/1K chars	—	—	$0.024/1K chars (Custom Neural)
Transcription	$0.01-0.02/req	$0.016/min	$0.024/min	$0.016/min
Voice ID (speaker verify / identify)	$0.0075/verification	N/A	$0.025/call (Connect Voice ID)	$0.01/txn (Speaker Recog. — Limited Access)

The only API that scores pronunciation at the phoneme level. Used by language learning apps serving thousands of students. No contracts — cancel anytime.

Try Speech AI — Zero Risk

Image Processing

You're paying $0.23 per image for background removal when you could pay $0.005. AI image processing at a fraction of the cost — background removal, upscaling, face restoration — production quality.

<5s per ImageNo File Size LimitsPNG Transparent Output

What you can do

Cut e-commerce photo costs 46xAutomate real estate listings at scaleScale visual content without designersRestore old photos automaticallyPrint-ready images in seconds

Endpoint	Brainiall	remove.bg	Let's Enhance	Cloud APIs
Bg Removal	$0.005/img	$0.23/img (HD)	N/A	~$0.01/img
Upscale 2x/4x	$0.003/img	N/A	$0.20/img	~$0.01/img
Face Restore	$0.005/img	N/A	N/A	N/A

46x cheaper than remove.bg. Same quality. No subscriptions. Pay only for what you use — $10 free to start.

Try Image APIs — No Subscription

Document AI

Stop wiring together a per-feature IDP pricing matrix. Turn a document image into structured data — doc-type-aware field extraction (receipt, invoice, ID, contract, form, generic), natural-language Q&A over a document, and table extraction — at one flat per-page price. Plus full-document PDF → Markdown.

One Per-Page Price6 Doc TypesPay Per Request

What you can do

Automate accounts-payable — invoices & receipts to JSONCapture ID-document fields for KYC onboardingPull parties, dates & obligations out of contractsDigitise paper forms into label/value pairsLift every table out of a report into headers & rowsAsk a document a direct question, get a cited answer

Endpoint	Brainiall	AWS Textract	Azure Doc Intelligence	Google Document AI
Field extraction (receipt / invoice / id / contract / form / generic)	$0.01/page	~$0.05–$0.10/page (forms / expense / ID)	~$0.01/page (prebuilt), ~$0.05 (custom)	~$0.065/page (specialized)
Document Q&A (natural-language question)	$0.01/page	~$0.015/page (Queries)	via custom models	via custom processors
Table extraction (headers + rows)	$0.012/page	~$0.015/page (Tables)	~$0.01/page (prebuilt layout)	~$0.03/page (Form Parser)
PDF → Markdown (full-document layout)	see PDF→Markdown	N/A	~$0.0015/page (Read)	~$0.0015/page (OCR)

One endpoint family, one per-page price — not a per-feature, per-document-type matrix. Self-serve from the first call. No minimum spend, no contract.

Try Document AI — No Contracts

Fraud & Risk

You're either training and hosting your own model on AWS Fraud Detector, stuck scoring only Stripe-processed payments with Radar, or paying a fraud platform a percentage of your GMV. Transactional & account fraud risk scoring as one flat per-event REST call — a 0-1 probability, the exact risk factors that drove it, and a recommended allow/review/deny decision — with a /feedback loop that re-calibrates to your data.

One Per-Event PriceExplainable Risk FactorsFeedback Loop

What you can do

Score every checkout/payment at authorization timeCatch fake-account & promo abuse at signupAccount-takeover signals from device / IP / geo mismatchesCard-testing detection — bursts across many cardsTune the allow / review / deny bands per requestFeed confirmed outcomes back — the model re-calibrates

Endpoint	Brainiall	AWS Fraud Detector	Stripe Radar	Sift / Kount / Signifyd
Fraud score (event → probability + risk factors + decision)	$0.0075/event	~$0.005–$0.03/prediction	~$0.05/transaction (Radar for Fraud Teams)	% of GMV / per-order contract
Feedback (report confirmed outcomes — re-calibrates)	free	via model retraining	n/a	n/a

One endpoint family, one per-event price — not a model you train, not a payment processor you're locked into, not a percentage of your GMV. Self-serve from the first call. No minimum spend, no contract.

Try Fraud Score — No Contracts

Brainiall Authenticity

Deepfake and AI-content detection is sold as an enterprise platform behind an annual contract (Reality Defender, Sensity) or bundled into a volume-priced moderation API (Hive); the hyperscalers ship nothing dedicated. AI-generated & manipulated media detection as one flat per-asset REST call — submit an image, video or audio asset, get a 0-1 likelihood, a verdict, and the exact forensic & provenance signals behind it — plus a /provenance endpoint that extracts embedded C2PA Content Credentials.

One Per-Asset PriceExplainable SignalsC2PA Provenance

What you can do

Flag AI-generated or edited ID photos before KYC passes themScreen user uploads for deepfake & synthetic contentVerify C2PA Content Credentials before publishing or citing mediaCatch AI-generated or manipulated insurance claim photosDetect AI-generated product imagery on marketplace listingsEvery verdict is explained by its signal list — no black box

Endpoint	Brainiall	Reality Defender	Hive AI	Sensity AI
AI-generated / deepfake detection (image, video, audio)	$0.01/asset	enterprise contract	~$0.003–$0.02/call	enterprise contract
C2PA Content Credentials extraction (/provenance)	$0.01/asset	partial	n/a	partial

One endpoint family, one per-asset price — not an enterprise platform behind a sales motion, not a moderation API where AI-detection is one label among many. Every verdict ships with the signals that drove it. Self-serve from the first call. No minimum spend, no contract.

Try Brainiall Authenticity — No Contracts

Brainiall Dubbing

Video dubbing is sold as a credit-based studio product (ElevenLabs Dubbing), a subscription localisation suite (Rask) or one feature of an avatar-video platform (HeyGen); the hyperscalers ship transcription, translation and speech synthesis as separate APIs but no end-to-end dubbing call. Brainiall dubs a video into another language as one async REST job — submit a video and a target language, poll the job, get a fully dubbed video back with the original timing preserved and a per-segment transcript.

Per-Minute PricingOriginal Timing PreservedFree Preview

What you can do

Localise marketing and launch videos into every market you sell inShip one training recording in the languages your learners speakDub product demos and release walkthroughs per regionRe-voice long-form video and podcasts for new-language audiencesPreview the translated transcript free before committing a renderEvery job returns a per-segment transcript — original and translation

Capability	Brainiall	ElevenLabs Dubbing	Rask AI	HeyGen
Video dubbing into another language	$0.30/min	~$1+/min equiv.	~$0.50–$1/min	subscription
Translated-transcript preview (no render)	free	credit-metered	plan-dependent	plan-dependent

One async REST job runs the whole pipeline — transcribe, translate, synthesize, time-fit, remux — priced per minute of video, not by credits or a subscription. The /preview endpoint and job-status polling are free; you only pay when you create a dub. Self-serve from the first call. No minimum spend, no contract.

Try Brainiall Dubbing — No Contracts

Brainiall Web

Web scraping is sold as a credit-metered crawl API (Firecrawl), a per-call HTML-fetch API (ScrapingBee), a compute-unit automation platform (Apify) or an enterprise data-collection suite (Bright Data); the hyperscalers ship no general web scraping or crawling API. Brainiall scrapes a URL into clean Markdown, crawls a site, maps its URLs or searches the web, one REST call each, billed per operation and ethical by construction.

Per-Operation PricingClean Markdown OutputEthical by Design

What you can do

Turn documentation sites and articles into clean Markdown for LLM and RAG pipelinesCrawl a whole site into a single searchable corpus, then keep it freshPull pricing, catalog and news pages on a schedule for market researchMap every URL of a site from its sitemap before a targeted crawlRe-scrape a page on a cadence and diff the Markdown to catch changesSearch the web for a query and get a ranked list of result pages

Capability	Brainiall	Firecrawl	ScrapingBee	Bright Data
Scrape a URL to clean Markdown	$0.002/op	~$0.003+/page	~$0.001-0.007/call	subscription
Honors robots.txt, never bypasses bot-detection	always	varies	varies	varies

One REST call runs the whole job: fetch, isolate the main content and convert to Markdown, priced per operation rather than by credits or a subscription. Scrape, crawl, map and search are billed at the same flat rate. Self-serve from the first call, no minimum spend, no contract. And it is ethical by construction: robots.txt and crawl-delay are honored and bot-detection is never bypassed.

Try Brainiall Web, No Contracts

Brainiall Interpreter

Speech-to-speech translation is shipped as a real-time streaming SDK inside a broad cloud speech service (Azure AI Speech), or not as a single API at all — Google and AWS make you chain translation and synthesis, or transcription plus translation plus synthesis, into a pipeline yourself. Brainiall translates a spoken clip into spoken audio in another language as one async REST job: submit a clip and a target language, poll the job, get the translated audio back with the source transcript and the translation.

Per-Job PricingSource + Translated TranscriptFree Job Polling

What you can do

Let support agents and callers who speak different languages hear each otherRe-voice announcements, messages and short clips into another languageTranslate a spoken voice message before you deliver itPair translated audio with both transcripts for captions and searchLet learners hear how a phrase sounds spoken in another languageEvery job returns the source transcript and the translation

Capability	Brainiall	Azure AI Speech	Google Cloud	AWS
Spoken clip translated into spoken audio	$0.04/job	~$2.50/audio-hr	chain 2 APIs	chain 3 APIs
Source + translated transcript returned	always	varies	assemble	assemble

One async REST job runs the whole pipeline — transcribe, translate, synthesize — priced per translation job, not per audio-hour or per character. Job-status polling is free; you only pay when you create a translation. Self-serve from the first call. No minimum spend, no contract.

Try Brainiall Interpreter, No Contracts

Brainiall Observability

LLM observability is sold as an open-source platform you self-host or its hosted cloud (Langfuse), as a proxy placed in front of your traffic (Helicone) or as tracing tied to one orchestration framework (LangSmith); the hyperscalers fold it into a broad application-monitoring suite. Brainiall ships a plain REST ingest API: POST a trace for every model call, GET aggregate latency, token, cost and error-rate stats, POST a heuristic eval. It is framework-agnostic, with nothing in your request path, billed per operation.

Per-Operation PricingNothing In Your Request PathFree Stats Polling

What you can do

Keep a searchable record of every LLM call and filter to the errored or slow onesRoll token counts and per-call cost into a running total per modelWatch p50, p95 and max latency over a window and alert on the tailScore model responses with heuristic checks in CI and fail the build on a regressionMeasure RAG groundedness by passing the retrieved context as the referenceReading aggregate stats is always free, so a dashboard can poll without burning quota

Capability	Brainiall	Langfuse	Helicone	LangSmith
Trace ingest + aggregate stats + evals	$0.0002/op	usage + seats	request + seats	~$0.0005/trace
Sits in your request path (adds latency)	never	no (SDK)	yes (proxy)	no (SDK)

One REST ingest API runs all three jobs (record a trace, aggregate the health, score a response), priced per operation rather than by usage units, requests or seats. It is framework-agnostic and never proxies your model traffic, so it cannot add latency or become a dependency of a live call. Reading aggregate stats is free; you only pay to ingest, query or evaluate. Self-serve from the first call, no minimum spend, no contract.

Try Brainiall Observability, No Contracts

Brainiall Agent Platform

Agent adoption is blocked by governance: teams can prototype quickly, but production buyers need tool permissions, policy checks, audit trails and a clear run history. Brainiall Agent Platform turns existing Brainiall APIs into governed workflows: create agents, approve tools, evaluate policy before a run, execute steps and keep every decision traceable.

Closed BetaPolicy ChecksAudit Trail

What you can do

Marketplace trust and safety agents for listings, reviews, messages and disputesSupport inbox agents that classify language, sentiment and PII before routingKnowledge assistants with governed retrieval and grounded answersDocument automation agents that route extracted fields into human reviewRegulated SaaS workflows with account-scoped runs and policy logsClosed beta access for teams that need production governance before broad rollout

Capability	Brainiall	DIY Orchestration	Agent Frameworks	Observability-only
Create governed agents and runs	yes	build yourself	app code	no
Policy checks before execution	built in	build yourself	varies	no
Tool permissions + run trace	built in	build yourself	partial	trace only

The beta gives enterprise and high-growth SaaS teams a governed way to adopt agents without exposing internal model choices or building auth, quotas, policy checks and run audit from scratch.

Request Agent Platform Beta

Bundles — Speech / NLP / Document / Safety Pro

Five feature bundles ship alongside existing Brainiall SKUs. No separate subscription — they share quotas with the underlying engines and unlock capabilities that AWS Polly, AWS Transcribe, Azure AI Speech, Azure AI Language, Azure AI Document Intelligence and Azure AI Content Safety sell as paid add-ons.

No separate tierShares underlying quotas+18 vertical doc types

What you can do

Speech Suite Pro — captions (SRT/VTT), audio-to-audio speech translation, transcript PII redaction, TTS speech marks, document translationNLP Pro — key phrases, aspect sentiment, zero-shot custom classifier, entity linking (public knowledge base), conversational PIIDocument AI Expansion — +5 prebuilt doc types (business card, W-2, health card, mortgage, pay stub), Markdown Layout, Skillsets enrichment, translation glossaryContent Safety Pro — Prompt Shields (jailbreak / injection), Groundedness Detection, Protected Material Detection, Multimodal Content UnderstandingDocument AI Verticals — 18 industry document schemas across Insurance (ACORD, claims, policy, loss runs), Healthcare (CMS-1500, EOB, superbills, prior auth, lab reports) and Finance & Tax (W-2, 1040, Schedule C, K-1, bank statements, balance sheets) packs

Bundle	Brainiall	AWS	Azure
Speech Suite Pro (5 endpoints)	bundled with Speech AI	Polly + Transcribe (separate)	AI Speech Captioning + Translation
NLP Pro (5 endpoints)	bundled with NLP Suite	Comprehend (per record)	AI Language (per record)
Document AI Expansion (3 endpoints + 5 doc types)	bundled with Document Intelligence	Textract Specialty (per page)	Doc Intelligence (per page)
Content Safety Pro (4 endpoints)	bundled with Text Intelligence	Bedrock Guardrails (limited)	AI Content Safety (per record)
Document AI Verticals (18 doc types)	bundled with Document Intelligence	Textract Queries (per field)	Custom model per type (~$50/1K pg)

Capabilities that take competitor APIs from "basic" to "useful" — bundled with existing Brainiall SKU usage, no add-on charge.

Explore the 5 bundles

One Vendor. Every API. Zero Risk.

NLP Suite, Speech AI, Image Processing, Document AI, Fraud & Risk, Observability, and governed Agent workflows — one API key, one bill, no contracts. Cancel anytime.

$10 free credits to start — no credit card required

Start Free

Volume tiers — committed-use discounts

For finance and procurement teams: prepaid annual commitments unlock 15-35% off the per-call price published on each product page. Tiers stack with the Marketplace + Stripe direct-bill paths. Email hello@brainiall.com for an enterprise quote and Enterprise MSA.

SKU	Product	Base / unit	Tier 1 (15% off)	Tier 2 (25% off)	Tier 3 (35% off)
S1	Brainiall Cutout (HD)	$0.05 / image	$0.0425 −15% ≥ 50K / mo	$0.0375 −25% ≥ 200K / mo	$0.0325 −35% ≥ 1M+ / mo
S2	Brainiall Audio Polish (Pro)	$0.08 / minute	$0.068 −15% ≥ 50K min / mo	$0.06 −25% ≥ 200K min / mo	$0.052 −35% ≥ 1M+ min / mo
S3	Brainiall Speaker ID	$0.06 / minute	$0.051 −15% ≥ 50K min / mo	$0.045 −25% ≥ 200K min / mo	$0.039 −35% ≥ 1M+ min / mo
S4	Brainiall Document Reader	$0.005 / page	$0.00425 −15% ≥ 50K pages / mo	$0.00375 −25% ≥ 200K pages / mo	$0.00325 −35% ≥ 1M+ pages / mo
S5	Brainiall Memory (events)	$0.5 / 1k events	$0.425 −15% ≥ 500K events / mo	$0.375 −25% ≥ 5M events / mo	$0.325 −35% ≥ 50M+ events / mo
S6	Brainiall Identity (Pro)	$0.01 / check	$0.0085 −15% ≥ 50K checks / mo	$0.0075 −25% ≥ 200K checks / mo	$0.0065 −35% ≥ 1M+ checks / mo
S7	Brainiall Content Guard (Fast)	$0.0008 / image	$0.00068 −15% ≥ 100K imgs / mo	$0.0006 −25% ≥ 500K imgs / mo	$0.00052 −35% ≥ 5M+ imgs / mo
S8	Brainiall OCR (Pro)	$0.005 / page	$0.00425 −15% ≥ 50K pages / mo	$0.00375 −25% ≥ 200K pages / mo	$0.00325 −35% ≥ 1M+ pages / mo
S9	Brainiall Vision Tagger (Standard)	$0.008 / image	$0.0068 −15% ≥ 50K imgs / mo	$0.0056 −30% ≥ 500K imgs / mo	$0.0048 −40% ≥ 5M+ imgs / mo

How committed-use pricing works

12-month annual commitment: prepay or quarterly invoice (NET-30 PO accepted at Tier 2+). Unused volume rolls over within the contract year.
Mix-and-match across SKUs: an annual contract can blend multiple products. Tier qualification is per-SKU based on monthly average usage.
Overage: above-tier usage billed at the same tiered rate (no surge penalty), measured monthly.
Auto-tiering: PAYG (Stripe / Marketplace) customers are automatically promoted to Tier 1 once they cross the threshold for 3 consecutive months — no contract required for Tier 1.
Enterprise (Tier 2+): also unlocks Enterprise MSA (liability cap up to $1M, governing-law options Delaware/NY/UK/Switzerland), 99.9% SLA, NET-30 PO, dedicated incident channel, custom data residency.
Free tier preserved: per-SKU free quotas (100 KYC checks, 500 imgs, 10 min audio, etc.) remain free regardless of commitment.

Pilot program: the first three customers per SKU willing to be publicly cited as references receive an additional 50% lifetime discount on top of Tier 3 (effective ~67% off list). See case studies / pilot program.