Switchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model selection, benchmarking, and cost/performance optimization.
Rust
docs: document API, model, and integration compatibility (#1153)
nachiketb-nvidia6ce4c29b14ff
diff --git a/docs/reference/compatibility.md b/docs/reference/compatibility.mdnew file mode 100644index 000000000..23c93bb3b--- /dev/null+++ b/docs/reference/compatibility.md@@ -0,0 +1,56 @@+# API and Model Compatibility++Check three things separately: the protocol format, the model's capabilities,+and the integration executing the algorithm. Supporting an API format does not+guarantee that a model supports every feature represented by that format.++## Supported API formats++These are the supplied adapter paths. Applications embedding libsy may own their+conversions instead. See **[Switchyard Protocols](../protocols/index.md)** for the+neutral types and integration flow.++| API | Supplied support | Read the exact contract |+|---|---|---|+| OpenAI Chat Completions | LLM request, buffered response, and streaming conversion; optional HTTP client. | [Requests](protocols/llm_requests.md), [responses](protocols/llm_responses.md) |+| OpenAI Responses | LLM request, buffered response, and streaming conversion; optional HTTP client. | [Requests](protocols/llm_requests.md), [responses](protocols/llm_responses.md) |+| Anthropic Messages | LLM request, buffered response, and streaming conversion; optional HTTP client. | [Requests](protocols/llm_requests.md), [responses](protocols/llm_responses.md) |+| System One | Typed Boolean, choice, and score questions through the supplied decision client. Used by open decision models, including Jev. | [Questions](protocols/decision_questions.md), [answers](protocols/decision_answers.md) |+| OpenAI Decisions | Typed decision questions through the supplied decision client. | [Questions](protocols/decision_questions.md), [answers](protocols/decision_answers.md) |++Provider fields have different equivalents and limitations across APIs.+Preservation supports lossless same-format replay. Cross-format conversion may+lose information. See **[translation policies and preservation](protocols/translation.md)**+before relying on a particular feature.++## Model requirements++Libsy algorithms select from the model IDs supplied by the host. There is no+fixed model-name allowlist in the routing core. Your endpoint and credentials+must serve the IDs you configure.++| Model role | Check before use | Guidance |+|---|---|---|+| Answer model | Context length, tool calling, input modalities, and requested output format for the actual workload. | [Model checks](../routing_algorithms/model_selection.md#check-the-efficient-model-before-you-route) |+| LLM judge | Structured output matching the algorithm's verdict schema and sufficient output budget. | [Judge compatibility](../routing_algorithms/llm_classifier_routing.md#judge-model-compatibility) |+| Decision model | Served question kinds and response fields matching the configured adapter. | [Decision protocol guide](../protocols/decisions.md) |+| Local or trained router | The trained assets and backend expected by that algorithm. | [Prefill routing](../routing_algorithms/prefill_routing.md) |++The **[measured model pairs](../routing_algorithms/model_selection.md#measured-pairs)**+and **[benchmark configurations](../benchmarks.md)** are examples with recorded+workloads and settings. They are not an exhaustive supported-model catalog.+Recheck capability and availability on the endpoint you will actually use.++## Integration restrictions++| Integration | Configuration owner | Important boundary |+|---|---|---|+| [Hosted OpenRouter router](https://openrouter.ai/docs/guides/routing/routers/switchyard-router) | OpenRouter request controls. | Only provider-exposed algorithms and settings apply; runner TOML is not its API. |+| [NeMo Relay](../integrations/nemo_relay.md) | Switchyard runner configuration inside Relay. | Check Relay version and plugin credential rules. |+| [LiteLLM](../../examples/litellm/README.md#supported-algorithm-semantics) | LiteLLM inventory and plugin configuration. | The experimental decision-only integration cannot serve intermediate model calls. |+| [Direct libsy](../reference/host_integration.md) | Your application. | Your host must fulfill every requested call and handle the final outcome. |+| [Local reference server](../operations/local_evaluation.md) | Runner TOML and local server settings. | Demo and personal use, not a supported production deployment. |++For dependency revisions, follow the **[quickstart](../getting_started.md#install)**+or the selected integration's instructions. Development callback examples use a+different API from the released `0.3.0` bindings and crates.diff --git a/mkdocs.yml b/mkdocs.ymlindex f1a75b190..33d7d3fac 100644--- a/mkdocs.yml+++ b/mkdocs.yml@@ -52,6 +52,7 @@ nav:- pi: integrations/pi.md- Oh My Pi: integrations/oh_my_pi.md- Reference:+ - API and Model Compatibility: reference/compatibility.md- Protocols:- LLM:- Requests: reference/protocols/llm_requests.md