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
Commits on main
refactor(server): move token counting out of algorithms (#316)
70aeb1f nachiketb-nvidia committed 3 months ago
fix(overflow): add hub GLM phrase to context-overflow detector (#314)
65491db Lin Jia committed 3 months ago
fix(translation): stop streams after in-band errors (#305)
a6aa645 Elyas Mehtabuddin committed 3 months ago
fix(server): render Anthropic error envelopes (#315)
e1895c7 nachiketb-nvidia committed 3 months ago
fix(libsy): keep classifier history from orphaning a tool result (#309)
702383e eric-liu-nvidia committed 3 months ago
fix(libsy): put judge system prompt in instructions, not messages (#312)
f4aa288 Lin Jia committed 3 months ago
docs(routing): the classifier judge is not shown client instructions (#311)
2a91472 ryan-lempka committed 3 months ago
fix(translation): report Responses max-token truncation as incomplete (#300)
5e83e2b ryan-lempka committed 3 months ago
fix(llm-class): fall back to capable when efficient overflows in escalation mode (#302)
e64c800 Lin Jia committed 3 months ago
docs: close RC1 reference gaps for [targets] and switchyard-server options (#295)
e687455 eric-liu-nvidia committed 3 months ago
feat(routing): add probability capability policy (#296)
5c25651 Ayush Agarwal committed 3 months ago
fix(server): handle graceful process shutdown (#294)
c1c1b41 nachiketb-nvidia committed 3 months ago
fix(packaging): read __version__ from installed metadata and pin documented install to Python 3.12 (#255)
38f2012 Elyas Mehtabuddin committed 3 months ago
feat(libsy): count classifier fail-open fallbacks on /metrics (#205)
ffca4e8 Elyas Mehtabuddin committed 3 months ago
fix(translation): reject malformed request fields (#287)
eea4e01 nachiketb-nvidia committed 3 months ago
fix(translation): preserve buffered Responses output (#284)
0e99eb4 nachiketb-nvidia committed 3 months ago
fix(translation): reduce Anthropic token fallback (#282)
a5fefa7 nachiketb-nvidia committed 3 months ago
fix(examples): forward full ensemble requests and cap concurrent explorations (#202)
8d47c2f Elyas Mehtabuddin committed 3 months ago
feat(server): report declared model capabilities on /v1/models (#200)
17b0e17 Elyas Mehtabuddin committed 3 months ago
fix(server): warn on duplicate target model ids (#180)
8928163 Elyas Mehtabuddin committed 3 months ago
feat(llm-classifier): custom schema support + multi target (>2) routing support (#275)
4ae99c2 Ayush Agarwal committed 3 months ago
fix(llm-client): enable Anthropic prompt caching by default (#233)
a9c04b3 nachiketb-nvidia committed 3 months ago
feat(llm-classifier): make system prompt configurable for user (#265)
14f75eb Ayush Agarwal committed 3 months ago