diff --git a/flake.lock b/flake.lock
index 9cf601fe..ec87d569 100644
--- a/flake.lock
+++ b/flake.lock
@@ -108,11 +108,11 @@
         "pre-commit-hooks": "pre-commit-hooks_3"
       },
       "locked": {
-        "lastModified": 1730277369,
-        "narHash": "sha256-yvQbeJbnnwCB68yv7uZXdGb+P7NMn5JMGBw0aBHymDI=",
+        "lastModified": 1732039290,
+        "narHash": "sha256-LQKY7bShf2H9kJouxa9ZspfdrulnZF9o4kLTqGqCDYM=",
         "owner": "nix-community",
         "repo": "crate2nix",
-        "rev": "151122427d030874ebef3517cda766a6984e6ed6",
+        "rev": "9ff208ce7f5a482272b1bcefbe363c772d7ff914",
         "type": "github"
       },
       "original": {
@@ -479,11 +479,11 @@
         "systems": "systems_6"
       },
       "locked": {
-        "lastModified": 1726560853,
-        "narHash": "sha256-X6rJYSESBVr3hBoH0WbKE5KvhPU5bloyZ2L4K60/fPQ=",
+        "lastModified": 1731533236,
+        "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=",
         "owner": "numtide",
         "repo": "flake-utils",
-        "rev": "c1dfcf08411b08f6b8615f7d8971a2bfa81d5e8a",
+        "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b",
         "type": "github"
       },
       "original": {
@@ -581,11 +581,11 @@
     },
     "nix-filter": {
       "locked": {
-        "lastModified": 1730207686,
-        "narHash": "sha256-SCHiL+1f7q9TAnxpasriP6fMarWE5H43t25F5/9e28I=",
+        "lastModified": 1731533336,
+        "narHash": "sha256-oRam5PS1vcrr5UPgALW0eo1m/5/pls27Z/pabHNy2Ms=",
         "owner": "numtide",
         "repo": "nix-filter",
-        "rev": "776e68c1d014c3adde193a18db9d738458cd2ba4",
+        "rev": "f7653272fd234696ae94229839a99b73c9ab7de0",
         "type": "github"
       },
       "original": {
@@ -853,11 +853,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1730687492,
-        "narHash": "sha256-xQVadjquBA/tFxDt5A55LJ1D1AvkVWsnrKC2o+pr8F4=",
+        "lastModified": 1732242723,
+        "narHash": "sha256-NWI8csIK0ujFlFuEXKnoc+7hWoCiEtINK9r48LUUMeU=",
         "owner": "oxalica",
         "repo": "rust-overlay",
-        "rev": "41814763a2c597755b0755dbe3e721367a5e420f",
+        "rev": "a229311fcb45b88a95fdfa5cecd8349c809a272a",
         "type": "github"
       },
       "original": {
@@ -978,11 +978,11 @@
         "nixpkgs": "nixpkgs_6"
       },
       "locked": {
-        "lastModified": 1732187990,
-        "narHash": "sha256-93xEH3aUs6+D5Kab9DGBUX9vrEpwhm839wdp2yCg9hI=",
+        "lastModified": 1732218602,
+        "narHash": "sha256-BElslL34KjOJCFMPkNtilOz6S/7iY7Vd72FNbRRWKDY=",
         "owner": "huggingface",
         "repo": "text-generation-inference-nix",
-        "rev": "f25a1cd889a6ae49c1e204232500005f82241a8b",
+        "rev": "f79638ac4e420e661321261744e745a3a747e182",
         "type": "github"
       },
       "original": {
diff --git a/server/poetry.lock b/server/poetry.lock
index 58450e77..fc8404c8 100644
--- a/server/poetry.lock
+++ b/server/poetry.lock
@@ -1288,12 +1288,12 @@ files = [
 
 [[package]]
 name = "marlin-kernels"
-version = "0.3.5"
+version = "0.3.6"
 description = "Marlin quantization kernels"
 optional = true
 python-versions = ">=3.7"
 files = [
-    {file = "marlin_kernels-0.3.5+cu123torch2.4-cp310-cp310-linux_x86_64.whl", hash = "sha256:58d4bf0aa1a9533acc05f1e5bf50f727ed0129848d1fa1feb2c5c3fa482518d4"},
+    {file = "marlin_kernels-0.3.6+cu123torch2.4-cp310-cp310-linux_x86_64.whl", hash = "sha256:afedaa9a15e8991442bc8c81f62833fbf5c1556ae9d7a5a9e13b747ce97beef9"},
 ]
 
 [package.dependencies]
@@ -1301,16 +1301,16 @@ torch = "*"
 
 [package.source]
 type = "url"
-url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp310-cp310-linux_x86_64.whl"
+url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp310-cp310-linux_x86_64.whl"
 
 [[package]]
 name = "marlin-kernels"
-version = "0.3.5"
+version = "0.3.6"
 description = "Marlin quantization kernels"
 optional = true
 python-versions = ">=3.7"
 files = [
-    {file = "marlin_kernels-0.3.5+cu123torch2.4-cp311-cp311-linux_x86_64.whl", hash = "sha256:a3a3653e6908db013ca96979a5ee1f6a8bb590ee7506a129e06b87d4a8cbb87d"},
+    {file = "marlin_kernels-0.3.6+cu123torch2.4-cp311-cp311-linux_x86_64.whl", hash = "sha256:c0c05621d5e87144415d8a6e439072bd844d5f3cb55e4c4c69eabdc4c94610f4"},
 ]
 
 [package.dependencies]
@@ -1318,16 +1318,16 @@ torch = "*"
 
 [package.source]
 type = "url"
-url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp311-cp311-linux_x86_64.whl"
+url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp311-cp311-linux_x86_64.whl"
 
 [[package]]
 name = "marlin-kernels"
-version = "0.3.5"
+version = "0.3.6"
 description = "Marlin quantization kernels"
 optional = true
 python-versions = ">=3.7"
 files = [
-    {file = "marlin_kernels-0.3.5+cu123torch2.4-cp312-cp312-linux_x86_64.whl", hash = "sha256:967b4765a591530a4b9160ae32f3f352a89ae4c71daf43220c99976987d76723"},
+    {file = "marlin_kernels-0.3.6+cu123torch2.4-cp312-cp312-linux_x86_64.whl", hash = "sha256:3be4662c8d25a3cdb1793dafe0e2e76dd600913a69a468e2c68d1fed4e149255"},
 ]
 
 [package.dependencies]
@@ -1335,16 +1335,16 @@ torch = "*"
 
 [package.source]
 type = "url"
-url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp312-cp312-linux_x86_64.whl"
+url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp312-cp312-linux_x86_64.whl"
 
 [[package]]
 name = "marlin-kernels"
-version = "0.3.5"
+version = "0.3.6"
 description = "Marlin quantization kernels"
 optional = true
 python-versions = ">=3.7"
 files = [
-    {file = "marlin_kernels-0.3.5+cu123torch2.4-cp39-cp39-linux_x86_64.whl", hash = "sha256:fbe607d5afd1e1fca6e294c3594a0ec279d1f9ea6a2fdf7f34ccb6180d15e195"},
+    {file = "marlin_kernels-0.3.6+cu123torch2.4-cp39-cp39-linux_x86_64.whl", hash = "sha256:89eac9d46bc084a256b538afda6053683eb7e505db0e0d4f6dbeca32368caac6"},
 ]
 
 [package.dependencies]
@@ -1352,7 +1352,7 @@ torch = "*"
 
 [package.source]
 type = "url"
-url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp39-cp39-linux_x86_64.whl"
+url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp39-cp39-linux_x86_64.whl"
 
 [[package]]
 name = "mdurl"
@@ -4066,4 +4066,4 @@ torch = ["torch"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.9,<3.13"
-content-hash = "7082f1983403ff58a1f0304e8bbf1197715b5156ddeea0f3e8287334d52c2617"
+content-hash = "995bb2a29e3e8cfa6ee8ddb139ec623c2cba4d7ad047066eb4952ccbab5579d5"
diff --git a/server/pyproject.toml b/server/pyproject.toml
index 3a47774f..d2263b1c 100644
--- a/server/pyproject.toml
+++ b/server/pyproject.toml
@@ -48,10 +48,10 @@ attention-kernels = [
   { url = "https://github.com/danieldk/attention-kernels/releases/download/v0.1.1/attention_kernels-0.1.1+cu123torch2.4-cp312-cp312-linux_x86_64.whl", python = "~3.12", optional = true },
 ]
 marlin-kernels = [
-  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp39-cp39-linux_x86_64.whl", python = "~3.9", optional = true },
-  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp310-cp310-linux_x86_64.whl", python = "~3.10", optional = true },
-  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp311-cp311-linux_x86_64.whl", python = "~3.11", optional = true },
-  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.5/marlin_kernels-0.3.5+cu123torch2.4-cp312-cp312-linux_x86_64.whl", python = "~3.12", optional = true },
+  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp39-cp39-linux_x86_64.whl", python = "~3.9", optional = true },
+  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp310-cp310-linux_x86_64.whl", python = "~3.10", optional = true },
+  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp311-cp311-linux_x86_64.whl", python = "~3.11", optional = true },
+  { url = "https://github.com/danieldk/marlin-kernels/releases/download/v0.3.6/marlin_kernels-0.3.6+cu123torch2.4-cp312-cp312-linux_x86_64.whl", python = "~3.12", optional = true },
 ]
 moe-kernels = [
   { url = "https://github.com/danieldk/moe-kernels/releases/download/v0.7.0/moe_kernels-0.7.0+cu123torch2.4-cp39-cp39-linux_x86_64.whl", python = "~3.9", optional = true },