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(changelog): highlight speculative and parallel workflows (#980)
nachiketb-nvidia19371a9fc3f7
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex c436d9e44..5ba34afc1 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -4,6 +4,21 @@ All notable changes to Switchyard are documented here. The format is based on[Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this projectadheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).+## [Unreleased]++### Added++- **Speculative workflows** - algorithms can run model and decision calls in+ parallel, stop one call, and let another continue. This includes decision+ models used as speculative tool callers. The same controls support general+ speculative and parallel workflows built on libsy. Cancellation stops local work;+ provider generation and billing may continue after disconnect.+ **Host API change (source-breaking):** Rust uses `call.respond(work).await?`;+ Python uses `await call.respond(work)` and `await call.fail(error)`. Pass work+ before awaiting it. See the+ [host integration guide](docs/getting_started.md#drive-the-algorithm).+ ([#948](https://github.com/NVIDIA-NeMo/Switchyard/pull/948))+## [0.3.0]Switchyard 0.3.0 builds on the native server and Rust library introduced in