Anthropic Provider

Rensei integrates with Anthropic Claude models. All recent Claude versions are supported, including reasoning, vision, and long-context variants.

Provider Summary

Attribute	Value
Provider ID	`claude`
Display name	Anthropic Claude
Config namespace	`anthropic`
Supported auth modes	`byok`, `metered`, `shared`, `host-session`
Requires endpoint	No
Category	Cloud

A model profile may include any subset of these modes in its auth_modes array. At dispatch time the resolver picks the first mode from that subset that the org and project access policy allows, in preference order byok > metered > shared > host-session. See Auth Modes for the full resolution algorithm.

Available Models

Rensei's catalog includes:

Claude Opus 4.1 (claude-opus-4-1-20250514) - Latest flagship; best for complex reasoning and agentic tasks. 200K context window.
Claude Sonnet 4 (claude-sonnet-4-20250514) - Balanced speed/quality; recommended for most workflows. 200K context window.
Claude Haiku (claude-3-5-haiku-20241022) - Fastest; ideal for simple tasks and cost-sensitive work. 200K context window.

Check rensei catalog list --provider claude to see the full current list.

Auth Modes

BYOK

Bring your own Anthropic API key:

Get an API key from console.anthropic.com.
In Settings → Integrations, click Add Provider and select Anthropic.
Paste your API key and click Test Connection.
In Settings → Model Profiles, create a profile with auth mode byok and select an Anthropic model.

Your org is billed directly by Anthropic.

Metered

Platform supplies a managed API key. Your org is billed through Rensei's cost ledger. Metered mode must be enabled for your organization by the platform operator - see Auth Modes for details.

Setup (user):

In Settings → Model Profiles, create a profile with auth mode metered and an Anthropic model.
Dispatch runs if your org is entitled; contact Rensei support to enable metered mode.

Shared

Free-tier platform key for trials. No setup required; quota is soft-limited per org per day.

In Settings → Model Profiles, create a profile with auth mode shared and an Anthropic model.
Dispatch runs if quota allows; returns SHARED_QUOTA_EXCEEDED otherwise.

Host-Session

Use OAuth credentials already logged in on your developer machine.

Run rensei setup and authenticate with Anthropic.
In Settings → Model Profiles, create a profile with auth mode host-session.
Dispatch from the same machine where you ran rensei setup.

Note: Only works with local capacity pools (providerId='local'). Dispatching to cloud sandboxes (e2b, Vercel, Modal) will fail.

Configuration

Context Window Override

By default, Rensei uses the model's native context window. Override at dispatch time:

{
  "providerConfig": {
    "anthropic": {
      "contextWindow": 100000
    }
  }
}

This is useful if you want to limit tokens for cost or latency reasons.

Prompt Caching

Claude supports prompt caching (cache_control = ephemeral or static). Rensei automatically enables it for compatible models. No explicit config required.

Vision

Claude Opus and Sonnet support vision. If your workflow includes image inputs, the provider automatically handles image encoding and sends them to Claude.

Reasoning Effort

Claude Opus 4.1 supports configurable reasoning effort via the thinking block:

low (fast) - Quick answers, minimal reasoning overhead.
medium (balanced) - Default; good balance of reasoning and speed.
high (deep) - Extended thinking; best for complex problems.

Set reasoning effort in the profile's Reasoning Effort dropdown or via API:

{
  "effort": "high"
}

Example Profile: Reasoning-Heavy Dispatch

rensei profile create \
  --name "anthropic-deep-reasoning" \
  --provider claude \
  --model-id "claude-opus-4-1-20250514" \
  --auth-modes byok,metered \
  --effort high \
  --scope project \
  --project-id my-project

The --auth-modes flag accepts a comma-separated list. The resolver will prefer byok when both are in the effective allowed set; if the project restricts to metered-only, metered is used automatically.

Then in a workflow LLM node:

nodeId: llm_call_1
nodeType: action/llm.inference
config:
  profileId: prof_anthropic_deep_reasoning
  systemPrompt: "You are an expert problem solver. Think deeply."

Pricing

View pricing in the model catalog:

rensei catalog show claude-opus-4-1-20250514 --format json | jq .pricing

Example:

{
  "inputPerMTokenCents": 3,
  "outputPerMTokenCents": 15
}

For metered mode, cost events are automatically emitted with the org ID for billing.

Troubleshooting

"No metered key is configured for provider 'claude'"

The platform's metered key pool for Anthropic is not configured. Contact your Rensei operator - deployment-level key pool setup is covered in the operator docs.

"Organisation is not entitled to use metered auth mode"

The org's metered_enabled feature flag is not set. Contact Rensei support to enable metered mode for your organization.

"AUTH_MODE_REQUIRES_LOCAL_CAPACITY"

You're dispatching a host-session profile to a cloud sandbox. Use a cloud-compatible auth mode (BYOK, metered, shared) or dispatch to a local pool.

Model not found in catalog

The model ID is correct but not in your catalog yet. Add it:

rensei catalog create \
  --provider claude \
  --model-id "claude-opus-4-1-20250514" \
  --display-name "Claude Opus 4.1"