← neout/agents · Docs · Anti-detect spec · Pricing

neout/agents vs browser-use cloud

An empirically-verified comparison. Every "they don't have" claim below was confirmed by direct API probes (their /api/v3/openapi.json) and hands-on UI testing — not marketing.

Methodology: we hit browser-use cloud's public OpenAPI (59 endpoints), ran 4 hands-on personas through their dashboard + playground, and probed their auth + error responses with curl. Findings are from 2026-06-09; we'll keep this page fresh as either platform ships changes.

API contract — empirically verified

Capability	neout/agents	browser-use cloud	Verified via
Error envelope	code · error · request_id	raw FastAPI `{detail: …}`	curl probe
`X-Request-Id` on every response	YES	NO	curl response headers
`Idempotency-Key` header	YES — 24h dedup, `Idempotent-Replay: true`	0 mentions in 59-endpoint OpenAPI	openapi.json grep
CORS preflight (`OPTIONS`)	204 + X-API-Key allowed	405 Method Not Allowed	curl OPTIONS
Auth header	`X-API-Key` OR `Authorization: Bearer`	`X-Browser-Use-API-Key` only	curl + spec
Public OpenAPI spec	YES — `/openapi.yaml`	YES — `/api/v3/openapi.json`	both fetched

Endpoints that exist on neout but NOT in their OpenAPI spec

The following are present in our /openapi.yaml but completely absent from browser-use cloud's /api/v3/openapi.json:

Endpoint	What it does
`POST /v1/webhooks`	Register a webhook subscription (browser-use cloud manages webhooks via dashboard only — no API)
`POST /v1/webhooks/{id}/ping`	Synchronously fire a test event to verify your handler
`GET /v1/webhooks/{id}/deliveries`	Per-delivery audit log with status_code + duration_ms + retry count
`GET /v1/webhooks/signing-secret`	HMAC-SHA256 secret for verifying delivery signatures
`POST /v1/keys` / `DELETE /v1/keys/{id}`	Programmatic API-key mint + revoke (theirs are dashboard-only)
`GET /v1/agents/{id}/fingerprint`	Wire-level UA + WebGL + canvas + tls preview — they don't expose any identity-quality signal
`GET /v1/agents/{id}/trust`	Per-factor trust-score breakdown — they have no trust score at all
`GET /v1/credits` + `/v1/credits/ledger`	Programmatic balance + per-job debit ledger
`POST /v1/quick/scrape`	One-shot synchronous scrape (no session lifecycle)
`POST /v1/quick/extract`	One-shot structured extraction (prompt or schema)
`POST /v1/quick/screenshot`	One-shot PNG capture
`POST /v1/quick/batch`	Fan-out N scrapes in one job
`GET /v1/health`	Liveness with build SHA

Identity quality — the category gap

browser-use cloud calls their persistent identities "Browser Profiles." Hands-on testing confirms they're cookie-storage jars with zero quality signal:

Signal	neout/agents Identity	browser-use Profile
Trust score (0–1000)	YES, with breakdown endpoint	None
Fingerprint preview	Full navigator + WebGL + canvas + TLS	None
Persona shaping at creation	OS, country, browser, UA, locale, persona name	Just "Profile Name" (string)
Proxy binding	Pinned to identity (sticky egress)	Decoupled — chosen per session
Last-used freshness	`last_session_at` + `session_count`	Only "cookie domains stored" text
Auth setup flow	Persistent + warmup recipes	"Launch a browser, log in yourself"

Live experience differences

Surface	neout/agents	browser-use cloud
Live agent preview	Interactive noVNC (you can take over the session)	Read-only `liveUrl` iframe
CDP-snippet UX	One-click "Connect ▾" → Python/TS/curl ready-to-paste	Bare CDP URL — devs hand-write boilerplate every time
Cost transparency on jobs	`llm_cost_usd` + `proxy_cost_usd` + `browser_cost_usd` + `total_cost_usd` on every job	Same shape in API; not surfaced on free-tier dashboard
Anti-detect transparency	Public per-OS fingerprint spec	"Advanced stealth" — opaque

What they do well (we acknowledge)

Per-session cost split surfaced in product UI (we have the schema; finishing the dashboard surface).
Smart routing for HTTP-fetch-only tasks — skips the browser, finishes in ~5s. We always provision a burner.
House-model menu (bu-mini/max/ultra + Gemini + GPT) one-click swap. We're BYOK-only.
codeMode / cacheScript — turns an agent run into a deterministic Playwright script. Genuine moat.
Workspaces + persistent file storage across sessions.

Try it

pip install neout-agents

from neout_agents import Client
c = Client(api_key="agt_live_...")

agent = c.create_agent(country="us", ttl="5m")
print(c.fingerprint(agent["id"]))      # Wire-level preview
print(c.trust_breakdown(agent["id"]))  # Why the score is what it is
print(c.connect(agent["id"]))          # Paste-ready Playwright snippet

Start free — 100 credits