Skip to content

Pull requests: llm-d/llm-d-router

New pull request New
76 Open 1,398 Closed
76 Open 1,398 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

feat(coordinator): async-broker step serving OpenAI compatible enqueue, wait, and passthrough modes kind/feature Categorizes issue or PR as related to a new feature. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files.
#2325 opened Aug 7, 2026 by BenjaminBraunDev Contributor Loading…
docs: add release-note fragment for PR #2321 size/XS Denotes a PR that changes 0-9 lines, ignoring generated files.
#2324 opened Aug 7, 2026 by llm-d-router-release-notes Bot Loading…
Add coordinator support for topology-aware P/D routing kind/feature Categorizes issue or PR as related to a new feature. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2319 opened Aug 7, 2026 by elevran Collaborator • Draft
7 tasks done
1
Skip body re-marshal/replace when no model-name rewrite occurred kind/cleanup size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files.
#2318 opened Aug 7, 2026 by elevran Collaborator • Draft
4 tasks done
1
tracing: honor OTLP TLS and auth environment variables kind/bug Categorizes issue or PR as related to a bug. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2312 opened Aug 6, 2026 by satyamg1620 Contributor Loading…
7 tasks done
1
fix(CrossReplicaSyncer): disentangle CrossReplicaSyncing interval from main ticket in PoolingDispatcher kind/bug Categorizes issue or PR as related to a bug. size/M Denotes a PR that changes 30-99 lines, ignoring generated files.
#2310 opened Aug 6, 2026 by Lucas-Fernandes-Martins Contributor Loading…
4
OSL-aware TokenLoadScorer kind/feature Categorizes issue or PR as related to a new feature. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2309 opened Aug 6, 2026 by DolevAdas • Draft
feat(epp): serve aggregated /v1/models from router kind/feature Categorizes issue or PR as related to a new feature. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2308 opened Aug 6, 2026 by zdtsw Member Loading…
1
3
test(coordinator): parallelize the e2e suite size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2302 opened Aug 6, 2026 by revit13 Contributor • Draft
Forward prefill response headers to decode requests size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2301 opened Aug 6, 2026 by hasB4K Contributor Loading…
4 tasks done
@elevran
2
tracing: configure SDK explicitly instead of mutating process env kind/cleanup size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2298 opened Aug 6, 2026 by satyamg1620 Contributor Loading…
7 tasks done
1
feat(scheduler): size approx-prefix-cache LRU to offload-extended KV cache capacity kind/bug Categorizes issue or PR as related to a bug. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files.
#2292 opened Aug 5, 2026 by BenjaminBraunDev Contributor • Draft
3 tasks done
1
3
fix: add dedicated APITypeMessages for Anthropic Messages API in routing sidecar kind/bug Categorizes issue or PR as related to a bug. size/M Denotes a PR that changes 30-99 lines, ignoring generated files.
#2289 opened Aug 5, 2026 by Gregory-Pereira Member • Draft
4
[ci]: Sign published container images with actions/attest kind/feature Categorizes issue or PR as related to a new feature. size/M Denotes a PR that changes 30-99 lines, ignoring generated files.
#2285 opened Aug 5, 2026 by hickeyma Contributor Loading…
3
feat(flowcontrol): split queue-wait budget by unavailability regime kind/feature Categorizes issue or PR as related to a new feature. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2284 opened Aug 5, 2026 by satyamg1620 Contributor Loading…
1
6
[wip][ci]: Add path based area/epp and area/dev PR labeling kind/feature Categorizes issue or PR as related to a new feature. size/M Denotes a PR that changes 30-99 lines, ignoring generated files.
#2281 opened Aug 5, 2026 by hickeyma Contributor • Draft
@elevran
7
[WIP] docs: add coordinator metrics specification size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2277 opened Aug 5, 2026 by revit13 Contributor Loading…
3
feat: add OpenTelemetry spans for tokenization and KV-cache event processing kind/feature Categorizes issue or PR as related to a new feature. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files.
#2272 opened Aug 4, 2026 by satyamg1620 Contributor Loading…
1
1
correlate logs with traces at span boundaries size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2271 opened Aug 4, 2026 by satyamg1620 Contributor Loading…
1
3
CICD: note SHA pin rationale for gosec SARIF upload action kind/documentation Categorizes issue or PR as related to documentation. size/XS Denotes a PR that changes 0-9 lines, ignoring generated files.
#2266 opened Aug 4, 2026 by elevran Collaborator Loading…
2
Propagate EPP-resolved priority to native backend schedulers (SGLang/… size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2265 opened Aug 4, 2026 by sywangyi Loading…
5
Peak prefill throughput size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files.
#2251 opened Jul 31, 2026 by azamikram Contributor • Draft
fix(observability): inject trace_id/span_id into request loggers kind/bug Categorizes issue or PR as related to a bug. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2248 opened Jul 31, 2026 by umakantv Loading…
2 tasks done
1
2
feat(kvcache): add opt-in xxhash hashing to the KV-block token processor kind/feature Categorizes issue or PR as related to a new feature. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2237 opened Jul 30, 2026 by vMaroon Member Loading…
2 tasks done
9
[WIP] Enable Envoy priorities for high availability for standalone mode kind/feature Categorizes issue or PR as related to a new feature. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.
#2228 opened Jul 29, 2026 by rlakhtakia Contributor Loading…
3
ProTip! Adding no:label will show everything without a label.