fix(inference): Nemotron Bolt downloads fail with 401 from prerelease repos

src/OpenClaw.Connection/LocalAi/LlamaServerRouterConfiguration.cs

@@ -34,7 +34,7 @@ public static LocalInferencePlan ResolveInstalledPlan(LocalAiResolvedInstall ins
         LocalAiInstallManifest manifest = install.Manifest;
         LlamaRuntimeVariant runtime = LlamaRuntimeCatalog.FindInstalled(manifest.RuntimeId)
             ?? throw new InvalidDataException("The managed llama-server runtime is no longer qualified.");
-        LocalModelInfo model = LocalModelCatalog.FindInstalled(manifest.ModelCatalogId)
+        LocalModelInfo model = LocalModelCatalog.FindInstalled(manifest.ModelCatalogId, manifest.ModelAsset.SourceUrl)
             ?? throw new InvalidDataException("The managed local AI model is no longer qualified.");
         return new LocalInferencePlan(runtime, model, ResolveQualifiedReceipt(manifest, runtime, model),
             LocalInferenceModelSelectionOrigin.Explicit, manifest.SelectedGpuId);

src/OpenClaw.Connection/LocalAi/LocalAiGatewayProviderDefinition.cs

@@ -141,7 +141,8 @@ public static string BuildPrimaryModel(LocalAiResolvedInstall install)
 
     private static LocalModelInfo GetQualifiedModel(LocalAiResolvedInstall install)
     {
-        LocalModelInfo model = LocalModelCatalog.FindInstalled(install.Manifest.ModelCatalogId)
+        LocalModelInfo model = LocalModelCatalog.FindInstalled(
+            install.Manifest.ModelCatalogId, install.Manifest.ModelAsset.SourceUrl)
             ?? throw new InvalidDataException("The managed Local AI model is no longer qualified.");
         if (!string.Equals(model.Id, install.Manifest.ModelAlias, StringComparison.Ordinal))
             throw new InvalidDataException("The managed Local AI model alias does not match the qualified catalog.");

src/OpenClaw.SetupEngine/LocalAiInstallReconciler.cs

@@ -304,7 +304,8 @@ public async Task<LocalAiReconcileResult> ReconcileAsync(
     /// <summary>Read-only inspection for onboarding; unlike reconciliation, never migrates or saves receipts.</summary>
     public async Task<bool> InspectAsync(LocalAiResolvedInstall install, CancellationToken ct)
     {
-        var model = LocalModelCatalog.FindInstalled(install.Manifest.ModelCatalogId);
+        var model = LocalModelCatalog.FindInstalled(
+            install.Manifest.ModelCatalogId, install.Manifest.ModelAsset.SourceUrl);
         var installedRuntime = LlamaRuntimeCatalog.FindInstalled(install.Manifest.RuntimeId);
         if (model is null || installedRuntime is null)
             return false;

src/OpenClaw.SetupEngine/LocalAiSetupSteps.cs

@@ -56,8 +56,9 @@ public override async Task<StepResult> ExecuteAsync(SetupContext ctx, Cancellati
                 StringComparison.OrdinalIgnoreCase)
                 ? LocalInferenceEligibility.EvaluateInstalled(hardware, selectedModelId)
                 : LocalInferenceEligibility.Evaluate(hardware, selectedModelId);
-        if (string.Equals(selectedModelId, LocalModelCatalog.NemotronBoltModelId, StringComparison.OrdinalIgnoreCase) &&
-            string.Equals(selectedModelId, ctx.Config.LocalAi.InstalledReceiptModelId, StringComparison.OrdinalIgnoreCase))
+        // Headless --config setup cannot carry the UI-only receipt hint. The
+        // stored receipt owns an installed Bolt recipe and its exact context.
+        if (string.Equals(selectedModelId, LocalModelCatalog.NemotronBoltModelId, StringComparison.OrdinalIgnoreCase))
         {
             try
             {

src/OpenClaw.Shared/Inference/Catalog/LlamaRuntimeCatalog.NemotronBolt.cs

@@ -4,13 +4,12 @@ namespace OpenClaw.Shared.Inference.Catalog;
 
 public static partial class LlamaRuntimeCatalog
 {
-    // Bolt needs LiLiCorr decoding from support-lilicorr plus per-layer expert loading.
-    // Pin the tested source and payload here while other models keep the stable runtime.
-    // Base: https://github.com/frozenblade1224/llama.cpp/tree/a972392d7db3678954c66f06744e535519314009
+    // Bolt uses upstream LiLiCorr support for the public drafter tensor names.
+    // Keep its runtime pin independent from the stable runtime used by other models.
     public const string NemotronBoltRuntimeFamily = "nemotron-bolt-lilicorr";
-    public const string NemotronBoltSourceCommitSha = "b3e4f6ce7f22adb7f1b1eaa012db8367d45a574a";
-    public const string NemotronBoltReleaseTag = "bolt-lilicorr-b3e4f6c";
-    public const string NemotronBoltArm64RuntimeId = "bolt-lilicorr-b3e4f6c-cuda13-arm64";
+    public const string NemotronBoltSourceCommitSha = "23f7aabc16c72f6f707dba227251c35d7bd1ed96";
+    public const string NemotronBoltReleaseTag = "b11700";
+    public const string NemotronBoltArm64RuntimeId = "bolt-lilicorr-b11700-cuda13-arm64";
 
     private static readonly LlamaRuntimeVariant s_nemotronBoltArm64 = new(
         NemotronBoltArm64RuntimeId,
@@ -19,12 +18,12 @@ public static partial class LlamaRuntimeCatalog
         Array.AsReadOnly(new[]
         {
             new PinnedArtifact(
-                "llama-bolt-lilicorr-b3e4f6c-cuda13-arm64",
+                "llama-bolt-lilicorr-b11700-cuda13-arm64",
                 ArtifactRole.RuntimeBinary,
-                new GitHubReleaseSource("joelagnel/llama.cpp", NemotronBoltReleaseTag, NemotronBoltSourceCommitSha),
-                "llama-bolt-lilicorr-b3e4f6c-bin-win-cuda-13.4-arm64.zip",
-                46_836_189,
-                new Sha256Digest("f2adf45fda36a336d2a539e418c6e26f51d50d6f7bb3c086c79771ea348a9ee5")),
+                new GitHubReleaseSource("ggml-org/llama.cpp", NemotronBoltReleaseTag, NemotronBoltSourceCommitSha),
+                "llama-b11700-bin-win-cuda-13.4-arm64.zip",
+                146_943_791,
+                new Sha256Digest("8474f9edb24e917e5bf0c3824dd79bda5659483ee6adfcbf27197d05defb2783")),
             new PinnedArtifact(
                 "cudart-b11320-cuda13-arm64",
                 ArtifactRole.RuntimeDependency,
@@ -41,15 +40,16 @@ public static partial class LlamaRuntimeCatalog
             RuntimeFile("cublas64_13.dll", 24_207_984, "49e8fa23d88ac0cbae27e200fc092b894dd55b4de5cb4293993986e29fb9b65a"),
             RuntimeFile("cublasLt64_13.dll", 193_128_560, "daf579ae36bb3c85e2340c57a1994f634e07c6f55ef70dd4cdc26695de21c752"),
             RuntimeFile("cudart64_13.dll", 606_832, "bd927ddf03823eeead7c8b261a8669d96764b89109311f4ab6accde8b7d97ec2"),
-            RuntimeFile("ggml-base.dll", 606_720, "a8e7cfb6f5ad7b866d08b5c012cf9469a9b61456f62d76137db00f25031ce4f4"),
-            RuntimeFile("ggml-cpu.dll", 667_136, "5e1fa7cd5d38f218714dd243e3f7079a01e11e7dd02c62490ab33eb5396b9b7a"),
-            RuntimeFile("ggml-cuda.dll", 54_052_864, "edf5fe4a86c17b0b64e16f2d634686f929133cfce8802dc7ab726d710afae0a4"),
-            RuntimeFile("ggml.dll", 49_664, "b2b8d88e1c1b9a95581a5690ddd5b1691099c2e6d5b7e1794185bac615a12c6d"),
-            RuntimeFile("llama-common.dll", 8_754_688, "c3f9a2d9756a222f359b14b21c0464d1fb2ba19ae5d77ee3333515086e454a83"),
-            RuntimeFile("llama-server-impl.dll", 3_414_016, "a2e42c8e93194fe67b463b708d7ec0e9f4c10fa2b4707b60126f50a07c463558"),
-            RuntimeFile("llama-server.exe", 10_240, "33eea374c24f120b9eaf2303d646f67af41d2db4b37e589276f86340e6efb4ca"),
-            RuntimeFile("llama.dll", 2_708_992, "a85d7984020e23437a0ea91adce0142b6b774925f3301c80dedd8b8961188bcd"),
-            RuntimeFile("mtmd.dll", 2_408_448, "1d12a069d7cb94bcca9e1d4647108a9bee8d3cd45bdc87623884c23888094f5b"),
-            RuntimeFile("vcomp140.dll", 367_160, "1a455c526e42d80d3aa3189e082500b6eb7837fcf8be0e9fbd1f26c92acdab71"),
+            RuntimeFile("ggml-base.dll", 667_136, "ad891be29b12d53c5e7f5b681b7ba9ebedb5101f04314d0cbd83795bcb460896"),
+            RuntimeFile("ggml-cpu.dll", 835_072, "e49583ceb5ad0c076f2b9905bac66eb1dd49d34fda14f0a4f4fd368fa2b9fef8"),
+            RuntimeFile("ggml-cuda.dll", 146_929_152, "e3c7a7b5a79cc71cb667957ad2e5b28644ea843344c217a31fde13ed22e1e8ae"),
+            RuntimeFile("ggml-rpc.dll", 215_040, "e7617fb63ae34218fbc15d4644057c4f44a3590afc21f8bfb4cabe6030996fc5"),
+            RuntimeFile("ggml.dll", 69_120, "cca7613b2124c1b2cd9e53663269cd64fbcb1b480a465872915130b967c50fec"),
+            RuntimeFile("llama-common.dll", 7_157_760, "b88408489b7445b4bba94cccf18d54b6f07618fae7a8a643c5db3d768f7fb5d6"),
+            RuntimeFile("llama-server-impl.dll", 8_471_040, "43ff8798b8bf397c6fd39cb47a0366a9a5fa0a72c17185e51261747451542349"),
+            RuntimeFile("llama-server.exe", 9_728, "acaea7451d287ee8812bc2f8175594cbb43008313450ecc22e6e2f4b69b6220d"),
+            RuntimeFile("llama.dll", 3_121_152, "bf1083da93ce656214a74cae8fdad3580ee96e0bd4e8b27e65c8b6b65c5d3005"),
+            RuntimeFile("mtmd.dll", 1_576_960, "4b34d7a35913a85a013b1050c143480c87c18aeb7f1421c0574447c2203270e4"),
+            RuntimeFile("libomp.dll", 764_928, "26caae17f29aaf2238f664375b663cd306d596bc9e36e780fa88356c51fe876a"),
         }));
 }

src/OpenClaw.Shared/Inference/Catalog/LocalModelCatalog.NemotronBolt.cs

@@ -2,7 +2,11 @@ namespace OpenClaw.Shared.Inference.Catalog;
 
 public static partial class LocalModelCatalog
 {
-    private static LocalModelInfo CreateNemotronBolt() => new(
+    // Existing receipts keep their exact prerelease pins until explicitly reinstalled.
+    // This recipe is never offered for fresh acquisition.
+    private static readonly LocalModelInfo s_prereleaseNemotronBolt = CreateNemotronBolt(prerelease: true);
+
+    private static LocalModelInfo CreateNemotronBolt(bool prerelease = false) => new(
         NemotronBoltModelId,
         "Nemotron 3.5 Bolt 74B-A9B (IQ2, LiLiCorr)",
         "Nemotron 3.5",
@@ -11,9 +15,10 @@ public static partial class LocalModelCatalog
             NemotronBoltModelId,
             ArtifactRole.ModelWeights,
             new HuggingFaceRevisionSource(
-                "NVIDIA-Nemotron/Nemotron-3.5-Bolt-74B-A9B-IQ2-GGUF",
-                "06c7fae4665f61a83319ad98e2b00a2824a5f390"),
-            "super3.5-vl.gguf",
+                prerelease ? "NVIDIA-Nemotron/Nemotron-3.5-Bolt-74B-A9B-IQ2-GGUF"
+                    : "nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-GGUF",
+                prerelease ? "06c7fae4665f61a83319ad98e2b00a2824a5f390" : "e0fecb69cb3e5e4c5e8a056fa2aec98c78e91246"),
+            prerelease ? "super3.5-vl.gguf" : "NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB.gguf",
             22_018_326_528,
             new Sha256Digest("91d63363e782c8a3d5f4be42816eba8e0d42dd2530bd07a7d0d2dc440599af75")),
         new LocalModelRunRecipe(
@@ -32,11 +37,13 @@ public static partial class LocalModelCatalog
                 "nemotron-3.5-bolt-lilicorr-nvfp4",
                 ArtifactRole.ModelWeights,
                 new HuggingFaceRevisionSource(
-                    "NVIDIA-Nemotron/Nemotron-3.5-Bolt-74B-A9B-IQ2-LiLiCorr-GGUF",
-                    "77844a4db96a62c0a6e14247af855b8ae8e2f89b"),
-                "originalLi5_final_step353859_NVFP4.gguf",
-                998_242_176,
-                new Sha256Digest("9c4033517096cfca23ad9450044186204ad1ab35efa77e058870f6f61fc245e0")),
+                    prerelease ? "NVIDIA-Nemotron/Nemotron-3.5-Bolt-74B-A9B-IQ2-LiLiCorr-GGUF"
+                        : "nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-LiLiCorr-GGUF",
+                    prerelease ? "77844a4db96a62c0a6e14247af855b8ae8e2f89b" : "182d2987939e12f458298b5d5f6ef98912f4ad6c"),
+                prerelease ? "originalLi5_final_step353859_NVFP4.gguf" : "NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-LiLiCorr.gguf",
+                prerelease ? 998_242_176 : 998_242_560,
+                new Sha256Digest(prerelease ? "9c4033517096cfca23ad9450044186204ad1ab35efa77e058870f6f61fc245e0"
+                    : "080e1ea41c9562a7804dbf30c6fbd4d17e216c6b8ff666f88e6edb1816010194")),
             draftAttentionLayerCount: 5,
             // Saved prompt snapshots and serialized recurrent checkpoints can
             // exceed the KV cache itself when agent conversations switch.

src/OpenClaw.Shared/Inference/Catalog/LocalModelCatalog.cs

@@ -525,6 +525,16 @@ public static IReadOnlyList<LocalInferenceRunProfile> GetProfiles(LocalModelInfo
             : s_legacyModels.SingleOrDefault(model =>
                 string.Equals(model.Id, id, StringComparison.OrdinalIgnoreCase)));
 
+    /// <summary>
+    /// Retains known earlier artifact pins under the same catalog id when the
+    /// receipt names their source URL. Full receipt validation is still required.
+    /// </summary>
+    public static LocalModelInfo? FindInstalled(string? id, string? modelSourceUrl) =>
+        string.Equals(id, NemotronBoltModelId, StringComparison.OrdinalIgnoreCase) &&
+        string.Equals(modelSourceUrl, s_prereleaseNemotronBolt.Weights.DownloadUri.AbsoluteUri, StringComparison.Ordinal)
+            ? s_prereleaseNemotronBolt
+            : FindInstalled(id);
+
     /// <summary>True when the id resolves only to a retired catalog entry.</summary>
     public static bool IsLegacy(string? id) => Find(id) is null && FindInstalled(id) is not null;
 

tests/OpenClaw.Connection.Tests/LocalAiBoltRuntimeTests.cs

@@ -52,11 +52,14 @@ public async Task BoltStartup_DoesNotDowngrade256KWhenOnly128KFits()
         Assert.Equal(262144, (await new LocalAiManifestStore(paths).LoadAsync())!.Manifest.ContextLength);
     }
 
-    [Fact]
-    public async Task BoltStartupAndRestart_Retain128KWhenMoreMemoryIsAvailable()
+    [Theory]
+    [InlineData(false)]
+    [InlineData(true)]
+    public async Task BoltStartupAndRestart_Retain128KWhenMoreMemoryIsAvailable(bool prerelease)
     {
         using var temp = new TempDirectory("bolt-runtime-retention-");
-        var paths = await PrepareBoltInstallAsync(temp, 131072);
+        var paths = await PrepareBoltInstallAsync(temp, 131072, prerelease);
+        var original = (await new LocalAiManifestStore(paths).LoadAsync())!.Manifest;
         var events = new SynchronizedEventLog();
         var platform = new FakePlatform();
         var hardware = new BoltHardwareProbe(48L << 30);
@@ -68,13 +71,41 @@ public async Task BoltStartupAndRestart_Retain128KWhenMoreMemoryIsAvailable()
         Assert.Contains("ctx-size = 131072", (await File.ReadAllTextAsync(paths.RouterPresetPath)).Split(Environment.NewLine));
         Assert.Equal(LocalAiRuntimeState.Healthy, (await runtime.RestartAsync()).State);
         Assert.Equal(2, hardware.Calls);
-        Assert.Equal(131072, (await new LocalAiManifestStore(paths).LoadAsync())!.Manifest.ContextLength);
+        var current = (await new LocalAiManifestStore(paths).LoadAsync())!.Manifest;
+        Assert.Equal(131072, current.ContextLength);
+        Assert.Equal(original.ModelId, current.ModelId);
+        Assert.Equal(original.ModelAsset, current.ModelAsset);
+        Assert.Equal(Assert.Single(original.AdditionalModelAssetsOrEmpty), Assert.Single(current.AdditionalModelAssetsOrEmpty));
+    }
+
+    [Fact]
+    public async Task PrereleaseBolt_RejectsPublicDraftReceipt()
+    {
+        using var temp = new TempDirectory("bolt-mixed-receipt-");
+        var paths = await PrepareBoltInstallAsync(temp, 131072, prerelease: true);
+        var install = (await new LocalAiManifestStore(paths).LoadAsync())!;
+        var draft = LocalModelCatalog.Find(LocalModelCatalog.NemotronBoltModelId)!.Recipe.DraftWeights!;
+        var mixed = install with
+        {
+            Manifest = install.Manifest with
+            {
+                AdditionalModelAssets = [new LocalAiAssetReceipt
+                {
+                    FileName = Path.GetFileName(draft.RelativePath), SourceUrl = draft.DownloadUri.AbsoluteUri,
+                    SizeBytes = draft.SizeBytes, Sha256 = draft.Sha256.Value,
+                }],
+            },
+        };
+
+        Assert.Throws<InvalidDataException>(() => LlamaServerRouterConfiguration.ResolveInstalledPlan(mixed));
     }
 
-    private static async Task<LocalAiPaths> PrepareBoltInstallAsync(TempDirectory temp, int context)
+    private static async Task<LocalAiPaths> PrepareBoltInstallAsync(TempDirectory temp, int context, bool prerelease = false)
     {
         var paths = new LocalAiPaths(temp.Path);
         var model = LocalModelCatalog.Find(LocalModelCatalog.NemotronBoltModelId)!;
+        var weights = prerelease ? NemotronBoltPrereleaseFixture.Weights : model.Weights;
+        var draft = prerelease ? NemotronBoltPrereleaseFixture.DraftWeights : model.Recipe.DraftWeights!;
         var runtime = LlamaRuntimeCatalog.Find(RuntimeArchitecture.Arm64, model.RuntimeFamily)!;
         var profile = LocalModelCatalog.GetProfiles(model).Single(p => p.ContextTokens == context);
         string cache = temp.Combine("hf");
@@ -92,17 +123,17 @@ string Snapshot(PinnedArtifact artifact)
             FileName = Path.GetFileName(artifact.RelativePath), SourceUrl = artifact.DownloadUri.AbsoluteUri,
             SizeBytes = artifact.SizeBytes, Sha256 = artifact.Sha256.Value,
         };
-        var source = (HuggingFaceRevisionSource)model.Weights.Source;
+        var source = (HuggingFaceRevisionSource)weights.Source;
         var manifest = ValidManifest() with
         {
             SchemaVersion = LocalAiInstallManifest.AdditionalAssetsSchemaVersion,
             EngineVersion = runtime.ReleaseTag, RuntimeId = runtime.Id, ModelCatalogId = model.Id,
             ModelAlias = model.Id, ModelId = source.RepositoryId + "@" + source.RevisionSha,
             RuntimeAssets = runtime.Artifacts.Select(Receipt).ToImmutableArray(),
-            ModelAsset = Receipt(model.Weights), ModelPath = Path.Combine("models", model.Weights.RelativePath),
-            ModelCacheRoot = cache, CachedModelPath = Snapshot(model.Weights),
-            AdditionalModelAssets = [Receipt(model.Recipe.DraftWeights!)],
-            AdditionalModelPaths = [Snapshot(model.Recipe.DraftWeights!)],
+            ModelAsset = Receipt(weights), ModelPath = Path.Combine("models", weights.RelativePath),
+            ModelCacheRoot = cache, CachedModelPath = Snapshot(weights),
+            AdditionalModelAssets = [Receipt(draft)],
+            AdditionalModelPaths = [Snapshot(draft)],
             ContextLength = context, KeyCachePrecision = profile.KeyCachePrecision,
             ValueCachePrecision = profile.ValueCachePrecision, DraftKeyCachePrecision = profile.DraftKeyCachePrecision,
             DraftValueCachePrecision = profile.DraftValueCachePrecision,