diff --git a/.changeset/bump_uniffi_size_budgets.md b/.changeset/bump_uniffi_size_budgets.md
new file mode 100644
index 000000000..6f9aa3995
--- /dev/null
+++ b/.changeset/bump_uniffi_size_budgets.md
@@ -0,0 +1,5 @@
+---
+livekit-uniffi: patch
+---
+
+Re-baseline the iOS and Android binary size budgets, which had fallen behind the shipping binaries and blocked the 0.1.10 release.
diff --git a/livekit-uniffi/Makefile.toml b/livekit-uniffi/Makefile.toml
index 3a9d357dc..90beb3c08 100644
--- a/livekit-uniffi/Makefile.toml
+++ b/livekit-uniffi/Makefile.toml
@@ -400,7 +400,7 @@ echo
 # are ~2× larger by construction (two archs in one Mach-O) and shouldn't be
 # the metric. Override the limit per release with SPM_SIZE_LIMIT_BYTES.
 [tasks.swift-check-size.env]
-SPM_SIZE_LIMIT_BYTES = { value = "1179648", condition = { env_not_set = ["SPM_SIZE_LIMIT_BYTES"] } }
+SPM_SIZE_LIMIT_BYTES = { value = "1638400", condition = { env_not_set = ["SPM_SIZE_LIMIT_BYTES"] } }
 
 [tasks.swift-check-size]
 private = true
@@ -543,13 +543,8 @@ run_task = "swift-package-flow"
 # Measured on the stripped .so inside the release AAR (AGP strips debug symbols
 # during assemble; the raw target/ artifact is larger and not what ships).
 # Override the limit per release with ANDROID_SIZE_LIMIT_BYTES.
-#
-# 1536 KiB as of data streams v2: the async foreign-trait machinery, the stream-closed and
-# open-stream-count surface, and trailer encryption checks put the arm64-v8a slice at 1490 KiB
-# (measured 2026-08; cargo's `strip = "symbols"` is already applied, so AGP's strip pass has
-# nothing further to remove).
 [tasks.android-check-size.env]
-ANDROID_SIZE_LIMIT_BYTES = { value = "1572864", condition = { env_not_set = ["ANDROID_SIZE_LIMIT_BYTES"] } }
+ANDROID_SIZE_LIMIT_BYTES = { value = "1769472", condition = { env_not_set = ["ANDROID_SIZE_LIMIT_BYTES"] } }
 
 [tasks.android-check-size]
 extend = "android-shared"