CareerCamp
2/4 warm
CAREER AI · 8 MODEL FAMILIES + AUTO ROUTER

One API surface.
Eight distinct signal types.

Text, voice, video frames, vectors, agent reasoning, a scored gauge, web lookups, a video timeline — CareerCamp doesn't wrap one generic model eight ways. Each family is its own instrument, self-hosted, independently documented.

Model families
8 shipped
Web intel sources
11 live 11 on our roadmap
CareerLM languages
English 100-language support on our roadmap
Salary coverage
PPP-modeled 196-country coverage on our roadmap

Channel console — click a channel to jump to its docs

MODEL_CHANNELS.active = 8 · signal preview, sandbox data

CSVM Auto — a router, not a 9th family

One endpoint that deterministically shape-matches an incoming request to one of the model families below — a keyword/shape heuristic, no extra model call spent deciding. Routes to 6 of 8 families.

Never routed: CareerWebAPI (no natural request shape) · CareerLM chat/stream (open-ended, no natural "done" point)

Every family, in full

Real spec fields — model tier, context window, format. What's not yet benchmarked is labeled that way, not hidden.

What's real today

Numbers here match what's actually deployed, not a target — this page updates as coverage grows.

WEB INTEL SOURCES
11 live
TTS / ASR BENCHMARKS
not yet published
LANGUAGES (CAREERLM)
English-primary
CAREERSCORE METHOD
deterministic sub-scores

Documentation

One page per family. Pick a channel on the left.

Productionhttps://careerstudiomax.com/api/camp/v1
Sandboxno sandbox tier yet — test against production with a low-limit key
AuthAuthorization: Bearer pk_live_YOUR_KEY

No dedicated CareerCamp SDK package is published in any language yet — every example below is a plain HTTP call, not an SDK import.

API Keys

Patch a key into the channels it needs. Nothing more.

Secret keys are shown once

When you generate a key below, its secret key, restricted key, and webhook secret appear one time only. Store it in a secrets manager — not a chat tool, screenshot, or shared doc. Lost it? Revoke and regenerate.

Generate a key

Safe to ship in frontend code.
Rate limit
Token limit

Fixed by your plan tier (starter: 60 req/min, 100k tok/min) — per-key adjustment isn't wired up yet.

Not enforced by the backend yet — the field is here for the future, not silently ignored without saying so.

Key types

Your keys

NameEnvChannelsRequestsStatus

Secret, restricted, and webhook keys were shown once at creation — not retrievable here.

Connections

Connections not yet built record your request as real, counted demand — instead of pretending to switch something on.

Web Intel actor catalogue

1 active connection, 11 real sources today. The list below also shows what's code-ready but inactive and what's on the roadmap — mixing them into one undifferentiated list is exactly how a "150+" claim becomes misleading.

Usage

Real per-key requests and wallet balance, exactly what GET /v1/usage returns for this account. No per-feature breakdown ("cover letters", "resume downloads") — that's a consumer-webapp concept with no equivalent tracking for API keys, so this doesn't invent one.

KeyEnvRequestsRPM limitMonthly spendLast used