terminal — blackbox agent boot
>> 30M+ DEVELOPERS · 400+ MODELS · MULTI-AGENT EXECUTION

_BLACKBOX.AI

The agent platform that runs anywhere. Dispatch Claude Code, Codex, and Blackbox in parallel — let the best solution win, automatically.

Developers
|
AI Models
|
VS Code Installs
|
IDEs Supported
$ blackbox guide --section 01

SURFACES
One platform, six places to work.

BLACKBOX AI meets you wherever you write code. Every surface dispatches to the same agent network — the same power, your preferred context.

01
CLI

Your terminal, supercharged. Dispatch competing agents from a single command. They analyze your codebase, generate solutions in parallel, and open PRs automatically.

Multi-agent parallel Auto PR creation CI/CD integration
$ blackbox agent run --task refactor-auth
[INIT] Loading codebase context...
[SCAN] 47 files analyzed in 1.2s
[DONE] PR #219 opened — 3 agents competed
02
IDE

Agents work inside Blackbox IDE or VS Code. Real-time code generation, context-aware refactoring, and integrated testing — right where you write code.

Inline completions Chat-driven edits Test runner built-in
// tab-to-accept completion
const result = await db.insert(tasks)
  .values({ title, userId, status }) Tab ↵
// ✓ context-aware — reads your schema
03
CLOUD

Deploy autonomous agents to the cloud. They monitor, fix, and optimize your codebase 24/7 — even while your team sleeps.

24/7 autonomous Auto monitoring Team dashboards
AGENT-CLOUD // running 24/7
[WATCH] Monitoring 3 repos...
[ALERT] Dep vulnerability in lodash@4.17
[FIX] Auto-patched. PR #312 ready.
04
API

OpenAI-compatible endpoints. Chat completions, multi-agent orchestration, and real-time WebSocket streaming. Drop in wherever you already use OpenAI.

OpenAI-compatible Multi-agent API WebSocket streaming
POST /v1/chat/completions
"model": "blackbox-pro",
"messages": [{"role":"user",...}]
// ✓ drop-in OpenAI replacement
05
MOBILE

Ship code from your pocket. Review agent work, approve PRs, and dispatch tasks from anywhere. Push notifications keep you in the loop.

Push notifications PR review + approve One-tap dispatch
📱 notification
Agent: PR #218 ready for review
Rate limiting middleware — 46/46 tests ✓
  [MERGE] [REQUEST CHANGES]
06
BUILDER

Describe it, agents build it. Go from natural language to a deployed full-stack application. Architecture, code, testing, and deployment — all automated.

NL → full-stack app Auto architecture One-click deploy
"Build a SaaS billing dashboard"
[GEN] Next.js + Stripe + Postgres
[TEST] 89 tests passing
[DEPLOY] live at yourapp.vercel.app ✓
$ blackbox guide --section 02

EXECUTE
Run agents in parallel.
Let the best solution win.

Dispatch the same task to Claude Code, Codex, and Blackbox simultaneously. The Chairman LLM evaluates every candidate and selects the winner — automatically.

// TASK Implement rate limiting middleware with Redis sliding window
BLACKBOX
Analyzing codebase...
CLAUDE CODE
Analyzing codebase...
CODEX
Analyzing codebase...
CHAIRMAN LLM — EVALUATING
CORRECTNESS
PERFORMANCE
RISK PROFILE
COMPLEXITY
WINNER BLACKBOX confidence: 0.94
PR #218 opened src/middleware/rate-limit.ts +47 -12 tests: 46/46 passing
Parallel Dispatch

Same task, all agents at once. No waiting, no picking — let the platform compete on your behalf.

Weighted Evaluation

Chairman LLM scores every candidate across correctness, performance, risk, and complexity — fully configurable.

Auto PR Creation

The winning solution is packaged into a PR with test results, diff, and evaluation breakdown. Merge in one click.

$ blackbox guide --section 03

MODELS
400+ models.
One subscription.

Every frontier model from every major lab — plus open-source alternatives — accessible through a single platform and a single API key.

Frontier Models xAI · Anthropic · OpenAI
Code-Optimized Codex · DeepSeek · StarCoder
Speed-Priority Minimax · Groq · Together
Voice Agents Real-time multimodal
Vision + Multimodal Image, video, code
model-selector — blackbox
FRONTIER
Grok-3 Claude 3.7 o3-mini GPT-4o Gemini 2.5 Claude Opus Claude Sonnet + many more
Included in all plans — $20–$80 monthly credits
CODE
Claude Code Codex DeepSeek Coder Blackbox-Code StarCoder 2 + many more
Used in multi-agent execution — compete for best output
FAST
Minimax-M2.5 Llama 3.3 Mixtral Qwen 2.5 Unlimited free on Minimax-M2.5
Unlimited free agent requests — blazing fast for routine tasks
VOICE
Voice Agent Real-time ASR Multimodal chat PRO plan+
Talk to your codebase — describe bugs, get live fixes
VISION
GPT-4V Gemini Vision Claude Vision Figma→Code PRO MAX+
Screenshot your UI → get production-ready component code
$ blackbox guide --section 04

INTEGRATE
4.2M installs.
#1 AI coding extension.

Works in VS Code, Cursor, JetBrains, and 35+ other IDEs. Or use the dedicated Blackbox IDE with agents built-in from the ground up.

rate-limit.ts api.ts auth.ts
12345 678910 1112131415
import { Redis } from '@upstash/redis';
import type { NextRequest } from 'next/server';
interface SlidingWindowConfig {
  maxRequests: number;
  windowMs: number;
  keyPrefix?: string;
}
export async function rateLimit(
  req: NextRequest,
  config: SlidingWindowConfig
) {
  const ip = req.headers.get('x-forwarded-for');  Tab ↵
}
▸ BLACKBOX /multi-agent
Refactor the rate limiter to use sliding window
01

Inline Completions

Context-aware multi-line suggestions drawn from your open files, imports, and project structure. Supports all major languages.

02

Chat-Driven Edits

Describe changes in natural language. Edits appear as inline diffs you review before accepting. Reference files with @path syntax.

03

Multi-Agent in Editor

Run /multi-agent from the chat panel. BLACKBOX, Claude Code, Codex, and Gemini compete. Chairman LLM picks the winner inline.

04

E2E Encryption

Your code never trains their models. End-to-end chat encryption on Pro Plus and above. Enterprise gets full training opt-out by default.

$ blackbox guide --section 05

BUILD
Describe it.
Agents ship it.

The Builder surface takes natural language and produces a fully deployed, production-ready application. Architecture, code, tests, and CI/CD — all included.

// USER PROMPT
Build a SaaS analytics dashboard with auth, billing via Stripe, and a live chart of user events
01
ARCHITECTURE
Next.js 15 + TypeScript + Tailwind · Postgres + Prisma · Stripe Billing · NextAuth
PENDING
02
CODE GENERATION
47 files · 4,200 lines · 18 API routes · 6 DB tables · Stripe webhooks
PENDING
03
TESTING
89 tests written · 89/89 passing · Coverage 91% · Playwright E2E added
PENDING
04
DEPLOY
yourapp.vercel.app · GitHub repo created · CI/CD pipeline active · DNS configured
PENDING

Full-Stack Generation

Frontend, backend, database schema, auth, payments — everything generated and wired together by the agents.

🎨

Figma to Code

Paste a Figma link or screenshot. Agents produce pixel-accurate, production-grade React components. (Pro Max+)

🔁

Iterative Editing

After generation, describe changes in natural language. Agents understand the entire generated codebase as context.

$ blackbox guide --section 06

SCALE
$2 to start.
Unlimited ceiling.

Every plan includes credits to frontier models. Start with PRO for $2 your first month — 80% off. No lock-in. Upgrade or cancel anytime.

PRO
$ 10 /month
80% OFF $2 first month
$20 worth of frontier model credits
  • All chat models (400+)
  • $20/mo model credits
  • Unlimited free Minimax-M2.5
  • Voice agent access
  • Multi-agent execution
  • App builder
  • Coding agent (35+ IDEs)
  • Slack integration
Get PRO
PRO MAX
$ 40 /month
$80/mo model credits
  • Everything in PRO PLUS
  • Unlimited agent requests
  • Figma to Code
  • Team collaboration features
  • Centralized billing & management
  • Advanced security controls
  • SAML SSO
  • Usage analytics & reporting
Get PRO MAX
ENTERPRISE

Training opt-out by default · Custom SLAs · On-premise deployment · Dedicated support · SAML SSO · Priority access with zero wait

Contact Sales →
$ blackbox guide --section 07

TRUST
30M developers.
Fortune 500 standard.

From indie hackers shipping solo products to engineers at the world's largest tech companies.

MicrosoftGoogleAmazonApple DeloitteAccentureSalesforceIntel OracleSAPCiscoPwC GitHubByteDanceInfosysCapgemini
★★★★★

"Multi-agent execution is a genuine game changer. I dispatch the same refactor task to three agents, Chairman picks the best diff, and I merge in 30 seconds. It's like having a team of senior engineers competing for me."

M
Marcus T.
Senior SWE · Verified user
★★★★★

"Switched from GitHub Copilot after seeing the 400+ model access. The fact that I can use Claude Code, o3-mini, and Blackbox's own model through the same interface and compare outputs side by side is unmatched."

S
Sarah K.
Full-Stack Dev · 4.2M installs community
★★★★★

"The Builder feature saved us 3 weeks. We described our MVP in plain English, Blackbox generated the full Next.js + Stripe + Postgres stack, all tests passed, and we were live on Vercel the same day."

A
Alex R.
Startup Founder · AI Unveiled community
$ blackbox guide --section 08

SHIP FASTER.
Stop waiting on reviews. Let agents compete.

Join 30M+ developers using BLACKBOX AI to build, refactor, test, and deploy faster than ever. Start free — your first month on PRO is $2.

✓ Free to start ✓ $2 first month on PRO ✓ 400+ models included ✓ Cancel anytime
agent-01 — final output