Release v0.13.1rc1

SkyPilot v0.13.1rc1

Get it now with:

uv pip install "skypilot>=0.13.1rc1"

---

What's Changed

• [Kubernetes] Surface remediation hint on abnormal->INIT cluster events by @SeungjinYang in #10060
• [Azure] Fix NIC provisioning broken by azure-mgmt-network 31.0.0 by @lloyd-brown in #10062
• [Dashboard] Truncate long managed job names in the jobs table by @stephenkevn in #10011
• [Dashboard] Add recovery count to managed job detail page by @stephenkevn in #10004
• [k8s] Guard against None pod.status.conditions in termination-reason lookup by @kevinmingtarja in #10065
• [Debug Dump] Resolve k8s coordinates from cluster YAML so INIT clusters still get their k8s/Kueue state dumped by @ishankaul1 in #10068
• [Test] Fix flaky test_helm_deploy_okta: tolerate current_url=None during dashboard redirect by @zpoint in #10073
• fix(runpod): correct and refresh region -> data center mapping by @ciprianf in #10072
• [Dashboard] Infra page: await sky check before reading enabled clouds on manual refresh by @cg505 in #9879
• [Client] Fix async get() not unwrapping the 'detail'-nested 500 payload by @cg505 in #10026
• Remove the API stream link from async request submission output by @claude[bot] in #10081
• [Jobs] Start consolidation-mode controllers only from the leader daemon by @cg505 in #10080
• [Test] chaos_proxy: jitter the kill interval to avoid phase-locked races by @cg505 in #10024
• [Jobs] Stream managed job logs on demand when a logging agent is configured by @aylei in #10069
• [k8s] Support storage.existingClaim to attach a pre-provisioned PVC by @aylei in #10085
• Add a pre-write validator hook for API server config saves by @kevinmingtarja in #10059
• [Dashboard] Render fenced code blocks in error messages by @SeungjinYang in #10088
• [Core] Extract external links from job logs server-side by @lloyd-brown in #9934
• [API] Grant private-workspace access to users added to allowed_users pre-login by @kevinmingtarja in #10079
• [API] Collect debug-dump request logs via LogProvider by @cg505 in #9852
• [Jobs] Give the controller process a single class identity by @cg505 in #10083
• [Dashboard] Add owner + activity toggles to Clusters page by @hentt30 in #9884
• [Dashboard] Default Clusters to My Clusters, add scope hint, unify toggle styling by @stephenkevn in #10089
• [Dashboard] Support admin-configured URL template external links by @lloyd-brown in #10090
• [Dashboard] Hint when a K8s context's nodes are filtered by allowed_nodes by @SeungjinYang in #10092
• Update the config doc and default value for rbac.restrict_config_to_admins by @DanielZhangQD in #10042
• [Kubernetes] Co-locate multi-node EFA jobs in one AZ by @trevanth24 in #10054
• [AWS] Skip unreachable me-south-1 region in catalog fetcher by @DanielZhangQD in #10095
• [k8s] storage.existingClaim activates the persistent-storage layout by @aylei in #10086
• [Test] Replace retired p3.2xlarge with g4dn.xlarge in tests by @aylei in #10098
• [Core] debug-dump: fast-fail unreachable kube contexts, parallelize per-cluster collection by @ishankaul1 in #10041
• [Jobs] Tolerate end-time fetch and log download failures in the controller by @kevinmingtarja in #10091
• [k8s] Avoid per-API-call urllib3 setLevel() to prevent logging fork deadlock on Python < 3.13 by @SeungjinYang in #10104
• [Kubernetes] Derive EFA count on scale-from-zero AWS clusters by @trevanth24 in #10053
• [Kubernetes] Cap provisioning thread pools at the number of work items by @SeungjinYang in #10111
• [Test] Replace retired p3/V100 AWS resources in examples by @SeungjinYang in #10107
• helm: make the persisted clients directory subPath configurable by @aylei in #10101
• [Test] Move zero-quota-failover test off retired AWS p3/V100 by @zpoint in #10113
• [Helm] Add apiService.host to control API server bind interface by @nicosuter in #10050
• [Test] Mark test_debug_dump_unreachable_context_fast_fail no_remote_server by @zpoint in #10116
• [Users] Allow specifying a role when creating a service account token by @DanielZhangQD in #10115
• [Test] Mark test_debug_dump_unreachable_context_fast_fail no_dependency by @zpoint in #10118
• [Docs] Update Sandboxes page with new features and current SDK syntax by @lloyd-brown in #10109
• [Jobs] Release the jobs-scheduler launch slot while the launch request is parked by @cg505 in #10013
• [Jobs] Emergency recovery: retry unexpected controller errors (reuse RECOVERING + recovery_source) by @cg505 in #9868
• [Azure] Support booting from private Shared Image Gallery images by @ahluwalij in #9843
• [k8s] Use disk_size instead of ephemeral_storage field by @kyuds in #9273
• [Core] Fix ray job submit failure in Ray cluster template by deconflicting dashboard agent port by @romilbhardwaj in #9856
• [k8s] Resolve urllib3 logger once at decoration time, not per API call by @SeungjinYang in #10108
• Cap provisioning thread pools at the number of work items(2) by @SeungjinYang in #10112
• [Metrics] Auto-detect local kubeconfig context to fix /gpu-metrics self-federation by @hentt30 in #10058
• [Test] Skip test_batch_diffusion on Azure (no L4 GPU in catalog) by @lloyd-brown in #10064
• [k8s] sky local up: add --num-nodes for multi-node local clusters by @zpoint in #10138
• [API] Add config post-save hook carrying old/new config and user by @ishankaul1 in #10133
• [Buildkite] Add concurrency_group marker for global per-test serialization by @zpoint in #10150
• [Core] Add SKYPILOT_DISABLE_LOCAL_API_SERVER to block local API server autostart by @Michaelvll in #10093
• [k8s] Request gzip-compressed API server responses by @aylei in #10153
• [Jobs] Clear SKYPILOT_DISABLE_LOCAL_API_SERVER when starting the controller's local API server by @zpoint in #10154
• [Kubernetes] Check ephemeral storage in check_instance_fits by @SeungjinYang in #10131
• Stash the yaml into the db during launch by @SeungjinYang in #10159
• [Metrics] Take local-context detection off the scrape path by @rohansonecha in #10175
• [Docs] Update SkyPilot logo and homepage architecture image by @stephenkevn in #10179
• [Kubernetes] - Support AWS Neuron (Trainium/Inferentia) accelerators by @trevanth24 in #10169
• [Test] Fix flaky test_pool_scale_down_with_job_count_priority by @zpoint in #10182
• [GCP] Add RTX PRO 6000 Blackwell (G4) support by @pauldonzier in #9634
• [API] Close SSH proxy websocket gracefully on cluster validation failure by @DanielZhangQD in #10183
• [Docs] Restyle Ask AI docs widget and add dark mode support by @stephenkevn in #10180
• [Docs] Update links, styling, and page organization by @romilbhardwaj in #10187
• [Docs] Remove demo card from docs homepage and README by @claude[bot] in #10189
• [AWS] Skip default-SG pre-creation for user-specified security groups by @rohansonecha in #10178
• [Dashboard] Use quicker count for the clusters empty-state CTA by @lloyd-brown in #10149
• [Dashboard] Update logo and favicon to new branding by @hopechong in #10181
• [Docs] Add Hugging Face Hub storage guide by @alex000kim in #10117
• [Dashboard] Render dynamically-registered columns on the Users table by @MadhavJivrajani in #10160
• [Test] Verify volume is actually mounted in test_volume_env_mount_kubernetes by @kevinmingtarja in #10190
• [CI] Tag the commit of each published nightly by @zpoint in #10192
• [Dashboard] Defer cluster fetch to background preload on the jobs page by @MadhavJivrajani in #10193
• [Nebius] Implement personalized pricing by @Jobarion in #9107
• [Dashboard] Add scope field to dashboard.external_links entries by @lloyd-brown in #10194
• [Jobs] Reduce jobs queue payload for field-filtered requests by @MadhavJivrajani in #10176
• [Core] Price region-free cost estimates with the joint regional minimum by @kevinmingtarja in #10199
• [Docs] Update Together AI in docs by @romilbhardwaj in #10198
• [Core] Surface concise health-probe failures in unhealthy-cluster events by @lloyd-brown in #10201
• [Dashboard] Show PENDING job reason in status tooltip by @lloyd-brown in #10202
• [LLM] slime on Job Groups: agentic coding-agent RL with a scalable inference fleet by @ishankaul1 in #10200
• [Volumes] Use SHORT queue for volume apply/delete to avoid LONG-pool starvation by @DanielZhangQD in #10195
• [Jobs] Add Prometheus gauges for managed-job recovery observability by @DanielZhangQD in #10186
• [API] Offload service-account auth DB lookups off the event loop by @kevinmingtarja in #10185
• [Test] Poll for async volume/PVC deletion in test_volumes_on_kubernetes by @zpoint in #10206
• [Kubernetes] OCI network_tier=best: GB200/GB300 NCCL profiles by @DanielZhangQD in #10204
• Release 0.13.1rc1 by @Michaelvll in #10219

New Contributors

• @nicosuter made their first contribution in #10050
• @ahluwalij made their first contribution in #9843
• @MadhavJivrajani made their first contribution in #10160

Full Changelog: v0.13.0rc1...v0.13.1rc1