Supported models

How the model catalog works, how to list it, and how new models roll out.

Dynoyard exposes a curated catalog of frontier OSS models behind one endpoint. You don’t manage providers or versions — you call a stable dynoyard-side slug and we route it to the best upstream.

The canonical list

The catalog evolves as new models ship, so the canonical source is always live, not this page:

GET /v1/models — the models enabled on your org, in OpenRouter-compatible shape (id, context length, pricing, capabilities).
Pricing page — the same catalog with per-token rates, billed per request from prepaid credits.

curl https://<your-org>.dynoyard.app/v1/models \
  -H "Authorization: Bearer sk-dyno-XXXX"

Families (mid-2026)

Family	Maker	Notes
Kimi	Moonshot	Long-context, thinking-mode reasoning
MiMo	Xiaomi	Reasoning, strong price/performance
DeepSeek	DeepSeek	R1-style reasoning
Qwen	Alibaba	Flagship `qwen-max`, fast `qwen-plus`, MoE variants
GLM	Zhipu	General + agentic

Exact ids and which are enabled on your org always come from /v1/models — don’t hardcode this table.

Capabilities

Each model in /v1/models carries a capabilities array so your SDK or router can light up the right features:

Capability	Meaning
`tools`	OpenAI-style function/tool calling
`streaming`	Server-Sent Events streaming
`reasoning`	Emits `reasoning_content` (thinking mode)
`vision`	Accepts image content blocks (provider-dependent)

This mirrors OpenRouter’s supported_parameters, so catalog-aware tooling (LiteLLM, openrouter-*) reads it without translation.

Slug stability & new models

Slugs are stable. Once a model id is published, your model field keeps working — we don’t rename underneath you.
New models appear automatically in /v1/models. No client code change, no new key, no new endpoint — switch by changing the model field.
Per-key restriction. Scope a key to a subset of the catalog with allowed models (see authentication).

Choosing a model per request

One key, one endpoint, switch models inline:

for model in ["kimi-k2-thinking", "mimo-v2-5-pro", "qwen-max-3-7"]:
    resp = client.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": "one-liner intro"}],
    )
    print(model, "→", resp.choices[0].message.content[:80])