Models

Syntic Code is model-agnostic. Pick the best model for the task, switch providers without changing your workflow, and configure fallbacks for resilience.

Supported Models

Model ID	Provider	Context	Best For
`kimi-k2-6`	Syntic (Moonshot K2.6 via Syntic proxy)	1M	Default — strong coding, large repos
`claude-opus-4-7`	Anthropic	1M	Heavy reasoning, complex refactors
`claude-sonnet-4-6`	Anthropic	200k	Balanced — fast and capable
`claude-haiku-4-5`	Anthropic	200k	Cheap, fast, simple tasks
`gpt-5`	OpenAI	256k	Alternative provider
`gemini-2.5-pro`	Google	2M	Massive-context tasks
`local/llama-3.1-70b`	Ollama / vLLM	Variable	Air-gapped deployments

Syntic Code is white-label friendly: the default kimi-k2-6 model is served through the Syntic proxy and branded as “Syntic AI” in product surfaces. Other providers can be enabled via your own API keys.

Setting the Default

syntic config set model.default kimi-k2-6

Or in settings.json:

{ "model": { "default": "kimi-k2-6" } }

Per-Session Override

syntic --model claude-sonnet-4-6

Or mid-session: /model claude-sonnet-4-6

Fallback Chains

Define a fallback in case the primary model is rate-limited or down:

{
  "model": {
    "default": "kimi-k2-6",
    "fallback": "claude-sonnet-4-6",
    "fallbackOn": ["rate_limit", "timeout", "5xx"]
  }
}

Provider Credentials

Each non-default provider needs an API key. Set via env var or syntic config set:

export ANTHROPIC_API_KEY=sk-ant-...
export OPENAI_API_KEY=sk-...
export GOOGLE_API_KEY=...
 
syntic config set providers.anthropic.apiKey "$ANTHROPIC_API_KEY"

Keys never leave your machine — they’re stored locally and sent directly to the provider.

Local Models

Run Llama, Qwen, or any OpenAI-compatible local model:

{
  "providers": {
    "local": {
      "baseUrl": "http://localhost:11434/v1",
      "apiKey": "ollama"
    }
  },
  "model": { "default": "local/llama-3.1-70b" }
}

Useful for air-gapped, regulated, or zero-egress environments.

Cost Tracking

syntic cost

Shows tokens, USD spent, and per-model breakdown for the current session or any saved session.

Settings — model and providers schemas
Enterprise — Centralized billing and quota management

Permissions & Security Enterprise