Model Registry

The Model Registry is a built-in, read-only catalog of known AI models and their providers. It provides capability data, tier classifications, and a pricing snapshot to help you create a shortlist. Provider capabilities, availability, and prices can change; confirm them with the provider and test the configured model before production use.

Tier system

Every model in the registry is classified into one of three tiers based on its capabilities:

Tier	Name	Description	Use case
Tier 1	Orchestrator	Cataloged for multi-step reasoning and tool use.	Supervisor agents and complex workflows to evaluate.
Tier 2	Sub-agent	Cataloged for focused tool-using work with lower latency or cost.	Specialist agents and bounded retrieval tasks to evaluate.
Tier 3	Utility	Lightweight models for simple operations. May not support tool calling.	Classification, summarization, data extraction.

Cataloged models

The provider column is catalog metadata, not the stored model type accepted by model create/import. For example, a catalog entry may say google, while a Gemini model configuration must currently use openai_compatible and Google’s compatible endpoint. See Model selection before creating a record.

Tier 1: Orchestrator

Model	Provider	Context	Max Output	Tools	Vision	Input $/1M	Output $/1M
Claude Opus 4.6	Anthropic	1M	32K	Yes	Yes	$5.00	$25.00
Claude Sonnet 4.6	Anthropic	1M	16K	Yes	Yes	$3.00	$15.00
GPT-5.4	OpenAI	272K	16K	Yes	Yes	$2.50	$15.00
GPT-5.2	OpenAI	200K	16K	Yes	Yes	$1.75	$14.00
Gemini 3.1 Pro	Google	1M	8K	Yes	Yes	$2.00	$12.00
Grok 4.1	xAI	2M	16K	Yes	Yes	$3.00	$15.00
DeepSeek V3.2	DeepSeek	128K	8K	Yes	No	$0.28	$0.42
GLM-5	Z.ai	200K	8K	Yes	Yes	$1.00	$3.20

Tier 2: Sub-agent

Model	Provider	Context	Max Output	Tools	Vision	Input $/1M	Output $/1M
GPT-5.4 Mini	OpenAI	128K	16K	Yes	Yes	$0.25	$2.00
Claude Haiku 4.5	Anthropic	200K	8K	Yes	Yes	$0.80	$4.00
Gemini 2.5 Flash	Google	1M	8K	Yes	Yes	$0.30	$2.50
GLM-4.7	Z.ai	128K	4K	Yes	No	$0.60	$2.20
Mistral Medium 3	Mistral	128K	8K	Yes	No	$0.40	$2.00

Tier 3: Utility

Model	Provider	Context	Max Output	Tools	Vision	Input $/1M	Output $/1M
GPT-5.4 Nano	OpenAI	128K	4K	No	No	$0.05	$0.40

API

List models

# All models
curl "$SYNTHETICBREW_URL/api/v1/models/registry"

# Filter by provider
curl "$SYNTHETICBREW_URL/api/v1/models/registry?provider=anthropic"

# Filter by tier (1 = Orchestrator, 2 = Sub-agent, 3 = Utility)
curl "$SYNTHETICBREW_URL/api/v1/models/registry?tier=1"

# Filter by tool calling support
curl "$SYNTHETICBREW_URL/api/v1/models/registry?supports_tools=true"

# Combine filters
curl "$SYNTHETICBREW_URL/api/v1/models/registry?provider=openai&tier=2"

Response

[
  {
    "id": "claude-sonnet-4-6",
    "display_name": "Claude Sonnet 4.6",
    "provider": "anthropic",
    "tier": 1,
    "context_window": 1000000,
    "max_output": 16000,
    "supports_tools": true,
    "supports_vision": true,
    "pricing_input": 3.0,
    "pricing_output": 15.0,
    "description": "Balanced reasoning model with full agent capabilities",
    "recommended_for": ["orchestrator", "sub_agent"]
  }
]

List providers

curl "$SYNTHETICBREW_URL/api/v1/models/registry/providers"

The provider catalog currently includes OpenAI, Anthropic, Google, Ollama, OpenAI Compatible, OpenRouter, Azure OpenAI, DeepSeek, Mistral, xAI, and Z.ai. These IDs are discovery metadata; only the stored provider types listed in the model-selection provider table are accepted when you create a model.

Admin Dashboard

The Admin Dashboard displays tier badges for stored model names that match a catalog entry:

Tier 1 models show a green badge — suitable for orchestrators.
Tier 2 models show a blue badge — suitable for sub-agents.
Tier 3 models show a gray badge — utility only.

Unrecognized names receive a Custom badge. The current Admin UI does not enforce or warn about role-to-tier matching, so verify tool-calling support before assigning a model to an orchestrator or tool-using specialist.