AI Agent Index
ByHeather MacAvelia·Independently reviewed·Published Mar 19, 2026·Updated Oct 25, 2026
Independently verified against live vendor data on Oct 25, 2026.

Enterprise-only coding agents from Blackbox: a CLI and an Agents API that run four agents on one task and let Chairman LLM pick the best result. Annual per-token commits.

How we scored it

Autonomy

4/5

Integrations

5/5

Pricing clarity

3/5

Evidence

4/5

Setup

2/5

The facts

Blackbox AI offers two coding agent products, Blackbox CLI and the Agents API, as part of a platform the company now positions as end-to-end encrypted AI inference for enterprises. In 2026 Blackbox moved to enterprise-only access. Its sign-in page tells existing users it is moving to enterprise-only access and offers refunds, the free and paid individual plans are no longer on its pricing page, and its privacy policy refers to the Visual Studio Code extension as discontinued.

Blackbox CLI is a terminal coding agent for macOS, Linux and Windows, installed through npm or a shell script. Plan mode produces an ordered list of steps for a task that a developer can approve, comment on or rewrite before any change applies. The /multi-agent command sends the same task to Blackbox, Claude Code, OpenAI Codex and Gemini CLI in parallel, each on its own Git worktree, and Chairman LLM evaluates the implementations and selects one, which the developer can override. The CLI reads AGENTS.md, plugins, hooks, skills and MCP servers from the repository, and a headless mode runs it in scripts and CI pipelines such as GitHub Actions and GitLab CI.

The Agents API exposes the same agents over HTTP. Each task runs in an isolated cloud sandbox with terminal, file system and Git access, and a run can stream logs, read and write files, take follow-up instructions and finish as a scoped pull request with a summary and test evidence. It works with GitHub, GitLab and Bitbucket, and Enterprise supports self-managed GitHub Enterprise Server, GitLab and Bitbucket Data Center instances with SSO.

Pricing is an annual enterprise commit through a purchase order, metered per token across 300+ models with no seat fees. Blackbox publishes per-model list rates in USD and quotes each contract against them, with discounts that grow with committed spend, and unused commitment expires at the end of the billing period. The Agents API is included with an Enterprise agreement, and Blackbox says buyers talk to sales from the start. Enterprise includes a dedicated forward-deployed engineer with implementation at no added cost, single-tenant deployment, data residency, SAML SSO, SCIM, RBAC and audit logs.

Blackbox says prompts and responses are not retained and are not used for training, and it routes away from model providers whose data policy it cannot verify. Requests are encrypted end to end and data at rest uses AES-256. Its trust center lists SOC 2, ISO 27001:2022 and ISO/IEC 42001:2023. Blackbox is a Microsoft Partner, with sign-in through Microsoft Entra ID, Okta and Google Workspace on Enterprise, and says more than 5M developers use its products. In July 2026 Artificial Analysis ranked Blackbox's NVIDIA Nemotron 3 Ultra deployment first on output speed.

Developers who want a terminal agent without an enterprise contract can look at Goose or Gemini CLI. Teams that want an autonomous cloud coding agent with self-serve plans can evaluate Devin ($20/mo) or Jules, and teams that work mainly inside an editor can look at Cursor ($8/mo).

Current state Q4 2026: Blackbox has moved to enterprise-only access, with coding agents delivered through Blackbox CLI and the Agents API under annual per-token commits. Individual plans and the Visual Studio Code extension are no longer offered, and the trust center lists SOC 2, ISO 27001 and ISO 42001.

Pricing

custom

View pricing ↗

Segment

Enterprise

Setup

moderate

Verified

Oct 25, 2026

Transparency

Partial

Contract

Annual Only

Data training

Not Trained

Human in loop

Optional

Blackbox AI pricing

Blackbox AI uses custom pricing: the vendor quotes a price rather than publishing one. Plans are billed annually only. Checked against the vendor's own pages on Oct 25, 2026.

Blackbox AI features

Agentic codingCode generationTerminal agentGit nativeAutonomous

Blackbox AI pros and limitations

Editorial assessment

Pros

  • ✓Multi-agent mode runs the same task through four coding agents in parallel, each on its own Git worktree or sandbox, and Chairman LLM selects the best implementation while the developer can override the choice.
  • ✓The Agents API covers the full run lifecycle over HTTP, from isolated cloud sandboxes and streamed logs to scoped pull requests on GitHub, GitLab or Bitbucket, including self-managed instances on Enterprise.
  • ✓Prompts and responses are not retained or used for training, requests are encrypted end to end, and Enterprise adds single-tenant deployment, data residency and a dedicated forward-deployed engineer.

Limitations

  • –Access is enterprise-only. There is no free or self-serve plan, every engagement starts with sales, and pricing is an annual per-token commit quoted against published list rates, with unused commitment expiring at the end of the billing period.
  • –The Visual Studio Code extension has been discontinued, and Blackbox's current product pages cover only the CLI and the Agents API, so developers who want editor-based assistance need another tool.
  • –Blackbox does not publish a setup time, its trust center lists SOC 2 without stating the report type, and most independent reviews on G2 and Product Hunt describe the earlier individual product.

Blackbox AI technical details

Deployment
cliapicloudweb
Model architectureMulti-model: routes across 300+ frontier and open-weight models, including open-weight models hosted by Blackbox and closed models from providers such as OpenAI, Anthropic and Google
Avg setup timeNot published by the vendor. Access is enterprise-only and starts with a sales-scoped commit, after which a forward-deployed engineer configures the account.
Autonomous rateAgents API tasks run autonomously in isolated cloud sandboxes and can finish as a scoped pull request. Multi-agent mode sends one task to four agents and Chairman LLM selects the result, with human override. The CLI's plan mode lets developers approve or edit steps before changes apply. Blackbox does not publish an autonomous completion rate.
Integrations
GitHubGitLabBitbucketDockerGitHub ActionsGitLab CIAWSGoogle CloudMicrosoft Entra IDOktaGoogle Workspace
Security
ISO 27001ISO 42001

Alternatives to Blackbox AI

Rating

4.1/ 5
★★★★★

Editorial score

How we score this →

Recognition

Listed 2026
Embed these on your site →

Built for

DevToolsEnterprise

Leave a review

Used Blackbox AI? Tell other buyers how it worked for you. Takes about two minutes.

Write a review

Compare

Related Content

Is this your product?

AI systems read this listing 23 times in the last 30 days. Last checked Oct 25, 2026. AI assistants repeat what this page says, so keep it current: Featured Listings are re-checked every 14 days.

Get Featured →Claim it free

AI Agent Price & Rating Tracker

Free, every two weeks. Price changes, new agent launches, acquisitions, and rating updates across the AI agent market. Verified against live vendor data, not vendor marketing.

No spam. Unsubscribe anytime. We never share your email. Read past issues