Introducing the Universal Managed Agents API.Read the announcement
Anthropic

Providers/Anthropic

Anthropic

Frontier lab behind the Claude family — the reference models for agentic coding and long-horizon work.

Operational
Model LabHarness BuilderFrontier

Models on Brainbase

Claude Fable

1 variant available1 new

claude-fable-5-1New

Claude Opus

5 variants available1 new

claude-opus-5New
claude-opus-4-8
claude-opus-4-7
claude-opus-4-6
claude-opus-4-5-20251101

Claude Sonnet

3 variants available1 new

claude-sonnet-5New
claude-sonnet-4-6
claude-sonnet-4-5-20250929

Claude Haiku

1 variant available

claude-haiku-4-5-20251001

Harnesses by Anthropic

Agent runtimes built by this vendor, managed end-to-end on Brainbase.

Claude Code
Platform-billed

Claude Code

claude_code

Anthropic's agentic coding harness, running on the Claude Agent SDK with full tool use, file editing, and sub-agent support.

Model delivery · Any Claude model, per task

About

Anthropic is the AI safety company behind Claude. Its Opus, Sonnet, and Haiku lines anchor most production agent deployments on Brainbase — Claude Sonnet is the platform's default model across open harnesses.

On Brainbase, Claude models run inside Anthropic's own harnesses — Claude Code and Claude Agent — or inside any open harness like OpenCode and Qwen Code. Every Claude model is vision-capable, so agents can read screenshots, diagrams, and documents out of the box.

Available on Brainbase:

  • Claude Fable: Anthropic's Mythos-class generation — its most intelligent models, including Claude Fable 5.1
  • Claude Opus: The flagship Claude line for the hardest agentic work, including Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Opus 4.6, Claude Opus 4.5
  • Claude Sonnet: The balanced Claude line — the default for most coding agents, including Claude Sonnet 5, Claude Sonnet 4.6, Claude Sonnet 4.5
  • Claude Haiku: Fast, inexpensive Claude for high-volume loops, including Claude Haiku 4.5

Run these models in Claude Code, OpenCode, or Qwen Code — selected per task like any other platform model.

Try it on Brainbase

One API call boots a managed sandbox on the claude_code harness pointed at claude-fable-5-1. Hosting, scaling, and billing are handled for you.

curl -X POST https://api.brainbaselabs.com/v2/agents \
  -H "Authorization: Bearer $BRAINBASE_API_KEY" \
  -d '{
    "title": "my-agent",
    "runtime_kind": "claude_code",
    "default_model": "claude-fable-5-1"
  }'
runtime_kind/claude_codemodel/claude-fable-5-1