Merge upstream main

.github/workflows/app.yml

@@ -9,6 +9,7 @@ on:
       - 'README.md'
       - '.github/workflows/app.yml'
   push:
+    branches: [main]
     paths:
       - 'app/**'
       - 'README.md'

.github/workflows/no-gpu.yml

@@ -13,10 +13,12 @@ on:
 permissions:
   contents: read
 
-# A pull request's newer push cancels its older run; every push to main keeps its own.
+# A pull request's newer push cancels its older run, and a newer push to main cancels main's older run: a landing train
+# pushes one commit a PR (seven runs in four minutes on 13 Nov), and the train's own run already tested the tree they
+# end at. The nightly schedule keeps a group of its own.
 concurrency:
-  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}
-  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
+  group: ${{ github.workflow }}-${{ github.event_name == 'schedule' && github.run_id || github.event.pull_request.number || github.ref }}
+  cancel-in-progress: ${{ github.event_name != 'schedule' }}
 
 jobs:
   benchmark-clients:
@@ -335,7 +337,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        suite: [fast, adapters, deepseek, lanes, deferred]
+        suite: [fast, adapters, native, deepseek, lanes, deferred]
     steps:
       - uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262 # v4.4.0
         with:
@@ -344,11 +346,29 @@ jobs:
         with:
           version: 0.17.0
           use-cache: false
+      - name: Metal toolchain for native engine host checks
+        if: matrix.suite == 'native'
+        run: xcrun -sdk macosx metal -v || { xcodebuild -downloadComponent MetalToolchain && xcrun -sdk macosx metal -v; }
       - name: Host checks and aggregate wall time
         shell: bash
         env:
           SUITE: ${{ matrix.suite }}
         run: |
           printf 'Runner class: %s %s\n' "$RUNNER_OS" "$RUNNER_ARCH"
+          if [ "$SUITE" = native ]; then
+            printf 'Native source: %s\n' "$GITHUB_SHA"
+            status=0
+            for phase in initial warm; do
+              TIMEFORMAT="test-native $phase wall seconds: %R"
+              if { time zig build test-host-ci-macos-native --test-timeout 30m --summary all; }; then
+                printf 'test-native %s exit: 0\n' "$phase"
+              else
+                code=$?
+                printf 'test-native %s exit: %s\n' "$phase" "$code"
+                status=$code
+              fi
+            done
+            exit "$status"
+          fi
           TIMEFORMAT='Host checks wall seconds: %R'
           time zig build "test-host-ci-macos-$SUITE" --test-timeout 30m

README.md

@@ -201,6 +201,7 @@ Conversions are explicitly marked: their upstream weights need the linked conver
 | Gemma 4 12B, converted MLX 4-bit | `hf download google/gemma-4-12B-it --revision 707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7 --local-dir ~/models/gemma-4-12b-hf`<br>`hf download google/gemma-4-12B-it-assistant --revision 46d4c6f13f0ac0ad827b915669b8df9b81c64c51 --local-dir ~/models/gemma-4-12b-assistant` | One Mac (M5 Ultra checked); converted text weights 6.3 GB, image weights 100 MB, plus assistant/caches. MLX 4-bit also served on one GB10. Pre-M5 Metal has a row path; the MTP assistant requires M5 tensor units. | [Convert and serve](docs/recipes/gemma-4-12b-native.md). Upstream download needs `hf`; conversion needs Python, mlx and mlx-lm. Serving is native; [CUDA qualification](CHANGELOG.md#114) includes the MTP assistant. |
 | Step-5-Preview, converted affine | `hf download stepfun-ai/Step-5-Preview --local-dir ~/models/step5-preview-fp8` | One 256 GB Mac; text 166.9 GiB, MTP adds 2.5 GiB and vision 3.7 GiB. | [Convert and serve](docs/recipes/step-5-preview.md). The documented 256 GB setup serves converted affine weights; converting the upstream FP8 download needs Python and mlx. |
 | MiniMax-M3.1-Flash, community affine 3/4-bit | `hf download BobClawblaw/MiniMax-M3.1-Flash-MLX-mix34 --revision 9524999057e0d277d26438606e033e2191fb465c --local-dir ~/models/MiniMax-M3.1-Flash-MLX-mix34`<br>`hf download BobClawblaw/MiniMax-M3.1-Flash-MLX-mix256k --revision 916e7ca1044f1c3e7566163096243c940327224d --local-dir ~/models/MiniMax-M3.1-Flash-MLX-mix256k` | One 256 GB M5 Ultra; 203.7 GiB (32,768-token window) or 189.2 GiB (147,456), plus caches. | [MiniMax-M3.1-Flash](docs/recipes/minimax-m3.1-flash.md). Set `MINIMAX_LOAD_LIMIT_GB`; images with `--vision`. Community checkpoints: lossy conversions of MiniMax's NVFP4 release (KL 0.244 / 0.295). MiniMax Community License. |
+| Recursive Flint 40B-A2.5B, MLX INT4 | `hf download RecursiveSI/flint-int4 --revision 3b93f9053b530488405ae9fca51d4900d6ce21f0 --local-dir ~/models/flint-int4` | One M5 Mac; M5 Ultra checked; 21.4 GiB resident plus caches. M1-M4 untested. | [Flint](docs/recipes/flint.md). Text and tools; always reasons; no draft head. |
 | Kolibri-1, block FP8 (research) | `tensorfold pull Aleph-Alpha/Kolibri-1@e52eb4627d11516b0c01de49210ab5a4e4061444` | One NVIDIA GPU; GB10 measured; 74 GB checkpoint. | [Kolibri](docs/recipes/kolibri-1.md). Text only; copy drafts, no EAGLE-3/neural drafter. |
 | Liquid AI d1-3B, decisions | Not documented yet. | Native CUDA; RAM minimum not published. | [API section below](#serve-flags). `/v1/systemone` and `/v1/decisions`. |
 | Fish Audio Drama 3, speech | Not documented yet. | Native Metal or CUDA; GB10 measured. | [Release audio instructions](CHANGELOG.md#110), [reference voices and dialogue speakers](docs/drama-http-voices.md). `/v1/audio/speech` with [sampling controls](docs/speech-sampling.md); checkpoint and `codec.pth` required. |

THIRD_PARTY_NOTICES.md

@@ -292,8 +292,7 @@ below); `tools/zig/bolt_ggml.py` reads them from that file. The IQ2_XXS and Q8_0
 Bolt's GGML checkpoints (`zig/kernels/cuda/ggml.cu`, `zig/src/cuda/ggml.zig`) are TensorFold's own.
 
 TensorFold ships no model weights. The `z-lab/Qwen3.8-27B-DFlash2` model card states Apache-2.0.
-The optional `incoai/GLM-5.3-Flash-DFlash2` model card states CC BY-NC-ND 4.0, for non-commercial use
-without derivatives; only the Python engine reads that drafter. Each checkpoint keeps its own license.
+The optional `incoai/GLM-5.3-Flash-DFlash2` model card states Apache-2.0 too. Each checkpoint keeps its own license.
 
 ## Structured output: libtfgrammar
 

build.zig

@@ -260,6 +260,9 @@ pub fn build(b: *std.Build) void {
     @import("zig/build/server.zig").cpuArtefacts(b, b.graph.host, optimize, build_options);
     @import("zig/build/mcdma_pair.zig").add(b, target);
     @import("zig/build/cache_index_timing.zig").step(b);
+    // the chat template's nesting, step, stack and memory limits (zig/src/core/template/limits_test.zig), on every host
+    const template_limits = b.createModule(.{ .root_source_file = b.path("zig/src/core/template/template.zig"), .target = b.graph.host, .optimize = .Debug, .link_libc = true });
+    b.step("test-template-limits-host", "Chat template nesting, render steps, stack and memory limits without a GPU").dependOn(&b.addRunArtifact(b.addTest(.{ .root_module = template_limits, .filters = &.{"security:"} })).step);
     sys_build.wire(b);
     @import("zig/build/checkpoint_config.zig").tests(b);
     @import("zig/build/checkpoint_config.zig").wire(b);