Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 0 additions & 4 deletions .lycheeignore
Original file line number Diff line number Diff line change
Expand Up @@ -60,10 +60,6 @@
# (afterwards lychee validates the URL again, catching any later rename/move).
^https://github\.com/ai-dynamo/dynamo/blob/main/deploy/inference-gateway/ext-proc/examples/onramp/agg\.yaml

# The v1.6.0 release entry lands with the current-release promotion; the tag
# URL 404s from CI until v1.6.0 is cut. Remove after the tag exists.
^https://github\.com/ai-dynamo/dynamo/releases/tag/v1\.6\.0$

# The classic dynamo-platform 1.6.0 chart publishes to helm.ngc.nvidia.com in a
# follow-up NGC publishing MR; the .tgz URL 404s until then. Remove once live.
^https://helm\.ngc\.nvidia\.com/nvidia/ai-dynamo/charts/dynamo-platform-1\.6\.0\.tgz$
Expand Down
12 changes: 7 additions & 5 deletions docs/fern/components/releases.data.ts
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ export interface Release {
}

export const CURRENT_VERSION = "v1.6.0";
export const CURRENT_DATE = "Oct 21, 2026";
export const CURRENT_DATE = "Oct 22, 2026";
export const CURRENT_TAG = "1.6.0";
export const CURRENT_WHEEL = "1.6.0";

Expand All @@ -87,7 +87,8 @@ const GH = "https://github.com/ai-dynamo/dynamo/releases/tag/";
export const RELEASES: Release[] = [
{
version: "v1.6.0",
date: "Oct 21, 2026",
notesHref: "/dynamo/dev/reference/releases/v1-6-0",
date: "Oct 22, 2026",
kind: "stable",
github: `${GH}v1.6.0`,
docs: "https://docs.nvidia.com/dynamo",
Expand All @@ -96,7 +97,7 @@ export const RELEASES: Release[] = [
delta:
"vLLM worker services renamed to worker, prefill and decode; Dynamo Operator no longer adopts resources it does not own and validates multinode settings more strictly; Grove v0.1.0-alpha.14 required for new multinode deployments, with Coherent updates now opt-in; AIConfigurator options renamed to AISimulate (--aic-* to --ais-*); Router policy YAML takes precedence over the scoring flags, which are deprecated with removal targeted for v1.7.0; structural tags on by default for supported parsers; TensorRT-LLM AutoDeploy backend removed and the TensorRT-LLM sidecar moved to the OpenEngine gRPC API; invalid DYN_REQUEST_PLANE and DYN_EVENT_PLANE values fail at startup.",
notesSummary:
"Pluggable flow control on one shared worker-selection core in the Router, TensorRT-LLM OpenEngine, LoRA and encode-prefill-decode support in the engine sidecars, V2 tool and reasoning parsers and a rerank endpoint in the Frontend, video-aware KV routing for SGLang, LPX components and role-specific pod templates on Kubernetes, Snapshot-backed engine failover, and runtime GPU budget reconfiguration in the Planner.",
"Pluggable flow control on one shared worker-selection core in the Router, V2 tool and reasoning parsers and a rerank endpoint in the Frontend, video-aware KV routing for SGLang, LPX components and role-specific pod templates on Kubernetes, Snapshot-backed engine failover, and runtime GPU budget reconfiguration in the Planner.",
},
{
version: "v1.5.1",
Expand Down Expand Up @@ -750,7 +751,7 @@ export const ARTIFACTS: Artifact[] = [
category: "container",
group: "component",
name: "dynamo-sidecar",
description: "Engine sidecars that run Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine",
description: "Dynamo Sidecar, which runs Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine",
meta: "CPU-only · AMD64/ARM64",
href: `${NGC_C}/dynamo-sidecar/tags`,
badge: "Experimental",
Expand Down Expand Up @@ -1425,8 +1426,9 @@ export interface ReleaseStats {
published 930/603/896, missing in both directions, so it cannot be trusted
to fill the rest. Leave them absent unless a method reproduces all three. */
export const RELEASE_STATS: Record<string, ReleaseStats> = {
"v1.6.0": { prs: 654, contributors: 148, firstTimers: 45, breaking: 41, knownIssues: 15 },
"v1.5.1": { breaking: 2, knownIssues: 2 },
"v1.5.0": { prs: 658, contributors: 123, breaking: 44, knownIssues: 12 },
"v1.5.0": { prs: 658, contributors: 123, firstTimers: 23, breaking: 44, knownIssues: 12 },
"v1.4.0": { prs: 640, contributors: 127, firstTimers: 29, breaking: 51, knownIssues: 19 },
"v1.3.0": { prs: 930, contributors: 125, firstTimers: 24, breaking: 24, knownIssues: 10 },
"v1.2.0": { prs: 603, contributors: 82, breaking: 5, knownIssues: 11 },
Expand Down
3 changes: 3 additions & 0 deletions docs/fern/index.yml
Original file line number Diff line number Diff line change
Expand Up @@ -1086,6 +1086,9 @@ navigation:
path: pages/reference/general/releases/release-history.mdx
collapsed: true
contents:
- page: Dynamo v1.6.0
path: pages/reference/general/releases/dynamo-v1-6-0.mdx
slug: v1-6-0
- page: Dynamo v1.5.0
path: pages/reference/general/releases/dynamo-v1-5-0.mdx
slug: v1-5-0
Expand Down
2 changes: 1 addition & 1 deletion docs/fern/pages/reference/general/release-artifacts.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -97,7 +97,7 @@ Current stable release: v1.6.0 (container tag `1.6.0`, wheel version `1.6.0`).
| container | dynamo-frontend | OpenAI-compatible API gateway with Endpoint Prediction Protocol (EPP) | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-frontend:1.6.0` |
| container | dynamo-planner | Standalone Planner used by Profiler jobs and Planner pods | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-planner:1.6.0` |
| container | kubernetes-operator | Operator that manages Dynamo deployments and CRDs | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/kubernetes-operator:1.6.0` |
| container | dynamo-sidecar (Experimental) | Engine sidecars that run Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine | CPU-only · AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-sidecar:1.6.0` |
| container | dynamo-sidecar (Experimental) | Dynamo Sidecar, which runs Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine | CPU-only · AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-sidecar:1.6.0` |
| wheel | ai-dynamo | Main package with backend integrations (vLLM, SGLang, TRT-LLM) | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install ai-dynamo==1.6.0` |
| wheel | ai-dynamo-runtime | Core Python bindings for the Dynamo runtime | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install ai-dynamo-runtime==1.6.0` |
| wheel | kvbm | KV Block Manager for disaggregated KV cache | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install kvbm==1.6.0` |
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -9,14 +9,14 @@ This page is a plain-markdown rendering of [`components/releases.data.ts`](https

{/* llms-tables:begin — generated by scripts/gen_llms_tables.py, do not edit */}

Current stable release: v1.6.0 (Oct 21, 2026; container tag `1.6.0`, wheel version `1.6.0`).
Current stable release: v1.6.0 (Oct 22, 2026; container tag `1.6.0`, wheel version `1.6.0`).

## Releases

| Version | Kind | Date | SGLang | TensorRT-LLM | vLLM | NIXL (SGL / TRT / vLLM) | UCX | Notes | Delta |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| main (ToT) | development head | - | 0.5.22 | 1.4.0rc0 | 0.31.0 | 1.5.0 / 1.3.1 / 1.4.1 | - | - | - |
| v1.6.0 | stable | Oct 21, 2026 | 0.5.21 | 1.3.0rc29 | 0.31.0 | 1.5.0 / 1.3.1 / 1.4.1 | 1.21.x | [release notes](https://github.com/ai-dynamo/dynamo/releases/tag/v1.6.0) | vLLM worker services renamed to worker, prefill and decode; Dynamo Operator no longer adopts resources it does not own and validates multinode settings more strictly; Grove v0.1.0-alpha.14 required for new multinode deployments, with Coherent updates now opt-in; AIConfigurator options renamed to AISimulate (--aic-* to --ais-*); Router policy YAML takes precedence over the scoring flags, which are deprecated with removal targeted for v1.7.0; structural tags on by default for supported parsers; TensorRT-LLM AutoDeploy backend removed and the TensorRT-LLM sidecar moved to the OpenEngine gRPC API; invalid DYN_REQUEST_PLANE and DYN_EVENT_PLANE values fail at startup. |
| v1.6.0 | stable | Oct 22, 2026 | 0.5.21 | 1.3.0rc29 | 0.31.0 | 1.5.0 / 1.3.1 / 1.4.1 | 1.21.x | [release notes](https://docs.nvidia.com/dynamo/dev/reference/releases/v1-6-0) | vLLM worker services renamed to worker, prefill and decode; Dynamo Operator no longer adopts resources it does not own and validates multinode settings more strictly; Grove v0.1.0-alpha.14 required for new multinode deployments, with Coherent updates now opt-in; AIConfigurator options renamed to AISimulate (--aic-* to --ais-*); Router policy YAML takes precedence over the scoring flags, which are deprecated with removal targeted for v1.7.0; structural tags on by default for supported parsers; TensorRT-LLM AutoDeploy backend removed and the TensorRT-LLM sidecar moved to the OpenEngine gRPC API; invalid DYN_REQUEST_PLANE and DYN_EVENT_PLANE values fail at startup. |
| v1.5.1 | patch | Oct 6, 2026 | 0.5.18 | 1.3.0rc25 | 0.28.0 | 1.4.0 / 1.3.1 / 1.3.2 | 1.21.x | [release notes](https://docs.nvidia.com/dynamo/dev/reference/releases/v1-5-0#v151) | Patch release. Fixes Router overload-hint expiry, min_tokens on tokenizer-free SGLang decode workers, split stop-sequence leaks, Qwen3-VL video routing across mixed workers, and reinforcement learning worker discovery on Kubernetes. Bounds client-supplied multimodal input (inline data: URL size, remote media downloads, image dimensions, base64 audio) and leaves aiohttp as the only HTTP client backend. Media fetches through an ambient HTTP proxy now require DYN_MM_TRUST_EGRESS_PROXY=1, and local SGLang diffusion input_reference files require DYN_MM_LOCAL_PATH. ModelExpress moves to v0.6.0. Backend pins are unchanged from v1.5.0. |
| v1.5.0 | stable | Sep 18, 2026 | 0.5.18 | 1.3.0rc25 | 0.28.0 | 1.4.0 / 1.3.1 / 1.3.2 | 1.21.x | [release notes](https://docs.nvidia.com/dynamo/dev/reference/releases/v1-5-0) | KVBM deprecated with removal targeted for v1.7.0; CRD storage version promoted to v1beta1 and the v1alpha1 admission webhook endpoints removed; Go EPP removed in favor of the Rust EPP shipped inside the Frontend image; Dynamo Snapshot moved to a standalone operator and chart, dropping the bundled snapshot chart, snapshot-agent image and DynamoCheckpoint CRD; AIConfigurator renamed to AISimulate; unified-backend entry point removed from the vLLM and SGLang workers; UCX 1.21.x. |
| v1.4.2 | patch | Aug 28, 2026 | 0.5.16 | 1.3.0rc22 | 0.26.0 | 1.3.0 / 1.3.1 / 1.3.2 | 1.21.x | [release notes](https://docs.nvidia.com/dynamo/dev/reference/releases/v1-4-0#v142) | Patch release and the first Dynamo Enterprise Support release: a curated set of release artifacts publishes under the -enterprise suffix on NGC, eligible for enterprise support, with no functional or binary differences from the open-source artifacts. Fixes NIXL loader-path resolution in the Frontend and SGLang Runtime images, removes the unused Nsight EFA metrics plugin, and tightens dependency pins (pillow v12.3.0 floor, plotext below v6, EFA Installer v1.50). Backend pins are unchanged from v1.4.0. |
Expand Down Expand Up @@ -53,7 +53,7 @@ Current stable release: v1.6.0 (Oct 21, 2026; container tag `1.6.0`, wheel versi

Release highlights (stable releases):

- v1.6.0: Pluggable flow control on one shared worker-selection core in the Router, TensorRT-LLM OpenEngine, LoRA and encode-prefill-decode support in the engine sidecars, V2 tool and reasoning parsers and a rerank endpoint in the Frontend, video-aware KV routing for SGLang, LPX components and role-specific pod templates on Kubernetes, Snapshot-backed engine failover, and runtime GPU budget reconfiguration in the Planner.
- v1.6.0: Pluggable flow control on one shared worker-selection core in the Router, V2 tool and reasoning parsers and a rerank endpoint in the Frontend, video-aware KV routing for SGLang, LPX components and role-specific pod templates on Kubernetes, Snapshot-backed engine failover, and runtime GPU budget reconfiguration in the Planner.
- v1.5.0: Pluggable worker selection policies and tiered KV indexing in the Router, guided tool-call streaming and request validation in the Frontend, image decoding at the Frontend for encode-prefill-decode deployments, v1beta1 CRD storage with the Rust EPP as the default Endpoint Picker, GPU Memory Service V1, and a standalone Snapshot operator.
- v1.4.0: Experimental cross-datacenter prefix routing and reservation replay in the Router, a vLLM-compatible generate token endpoint, tokenizer L1 prefix cache on by default, NIXL disaggregation for vLLM-Omni pipelines, and the Spica deployment simulator.
- v1.3.0: Tool-calling and reasoning overhaul, RL rollout serving, the largest Router buildout to date, SLA-driven Planner autoscaling, and production GPU Memory Service on Kubernetes.
Expand Down Expand Up @@ -181,7 +181,7 @@ Release highlights (stable releases):
| container | dynamo-frontend | OpenAI-compatible API gateway with Endpoint Prediction Protocol (EPP) | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-frontend:1.6.0` |
| container | dynamo-planner | Standalone Planner used by Profiler jobs and Planner pods | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-planner:1.6.0` |
| container | kubernetes-operator | Operator that manages Dynamo deployments and CRDs | AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/kubernetes-operator:1.6.0` |
| container | dynamo-sidecar (Experimental) | Engine sidecars that run Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine | CPU-only · AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-sidecar:1.6.0` |
| container | dynamo-sidecar (Experimental) | Dynamo Sidecar, which runs Dynamo beside a stock SGLang, TensorRT-LLM, or vLLM engine | CPU-only · AMD64/ARM64 | `nvcr.io/nvidia/ai-dynamo/dynamo-sidecar:1.6.0` |
| wheel | ai-dynamo | Main package with backend integrations (vLLM, SGLang, TRT-LLM) | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install ai-dynamo==1.6.0` |
| wheel | ai-dynamo-runtime | Core Python bindings for the Dynamo runtime | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install ai-dynamo-runtime==1.6.0` |
| wheel | kvbm | KV Block Manager for disaggregated KV cache | Python 3.10–3.12 · Linux (glibc v2.28+) | `uv pip install kvbm==1.6.0` |
Expand Down Expand Up @@ -275,8 +275,9 @@ Release highlights (stable releases):

| Release | PRs | Contributors | First-time contributors | Breaking changes | Known issues |
| --- | --- | --- | --- | --- | --- |
| v1.6.0 | 654 | 148 | 45 | 41 | 15 |
| v1.5.1 | - | - | - | 2 | 2 |
| v1.5.0 | 658 | 123 | - | 44 | 12 |
| v1.5.0 | 658 | 123 | 23 | 44 | 12 |
| v1.4.0 | 640 | 127 | 29 | 51 | 19 |
| v1.3.0 | 930 | 125 | 24 | 24 | 10 |
| v1.2.0 | 603 | 82 | - | 5 | 11 |
Expand Down
Loading
Loading