Launch bounded immutable GGUF inference on free CPU

Public cpu-basic Docker Space. Exact Q4_K_M artifact is preloaded at an immutable revision, rehashed before load, and served with strict input/output/concurrency limits plus health and identity endpoints. No secrets or paid hardware.

.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.pytest_cache
+local-evidence
+*.gguf

Dockerfile

@@ -0,0 +1,25 @@
+FROM python:3.11-slim-bookworm@sha256:b18992999dbe963a45a8a4da40ac2b1975be1a776d939d098c647482bcad5cba
+
+ENV PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    PIP_DISABLE_PIP_VERSION_CHECK=1 \
+    HF_HUB_DISABLE_TELEMETRY=1 \
+    DO_NOT_TRACK=1 \
+    HF_HUB_OFFLINE=1 \
+    HOME=/home/user \
+    PATH=/home/user/.local/bin:${PATH}
+
+RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates libgomp1 \
+    && rm -rf /var/lib/apt/lists/* \
+    && useradd --create-home --uid 1000 user
+USER user
+WORKDIR /home/user/app
+
+COPY --chown=user:user requirements.txt ./requirements.txt
+RUN python -m pip install --no-cache-dir --user -r requirements.txt
+COPY --chown=user:user . .
+
+EXPOSE 7860
+HEALTHCHECK --interval=30s --timeout=5s --start-period=120s --retries=3 \
+  CMD python -c "import urllib.request; urllib.request.urlopen('http://127.0.0.1:7860/health', timeout=4)" || exit 1
+CMD ["python", "-m", "uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860", "--workers", "1", "--limit-concurrency", "16", "--timeout-keep-alive", "5", "--no-server-header"]

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 SZL HOLDINGS LLC
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -1,10 +1,61 @@
 ---
-title: Szl Model Inference Lab
-emoji: 🦀
-colorFrom: green
-colorTo: gray
+title: SZL Model Inference Lab
+emoji: 🧪
+colorFrom: indigo
+colorTo: blue
 sdk: docker
-pinned: false
+app_port: 7860
+license: apache-2.0
+short_description: Bounded GGUF inference on free CPU, immutable identity.
+models:
+  - SZLHOLDINGS/SZL-Khipu-1.5B-GGUF
+tags:
+  - gguf
+  - llama.cpp
+  - cpu
+  - provenance
+  - bounded-inference
+suggested_hardware: cpu-basic
+startup_duration_timeout: 30m
+preload_from_hub:
+  - SZLHOLDINGS/SZL-Khipu-1.5B-GGUF SZL-Khipu-1.5B-Q4_K_M.gguf,training_receipt.signed.json,eval_receipt.signed.json,owner_pubkey.json 67d60ec577730747055491640cfb91fc4a4b5d25
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# SZL Model Inference Lab
+
+A public, zero-secret, bounded CPU demonstration for the exact
+`SZLHOLDINGS/SZL-Khipu-1.5B-Q4_K_M.gguf` bytes at immutable model commit
+`67d60ec577730747055491640cfb91fc4a4b5d25`.
+
+The runtime verifies the 986,047,904-byte file against SHA-256
+`13c1a1993063e1dff92f7413ccf48eaca6d48efc8801ae9af35961ae3396623a`
+before loading it. It requires no provider token or Space secret and is intended
+for the Hub's free `cpu-basic` hardware only.
+
+## Boundaries
+
+- One inference at a time; excess concurrent calls receive HTTP 429.
+- 1,200 input characters, 32 generated tokens, and a 45-second generation budget.
+- Greedy decoding (`temperature=0`); outputs are model-generated and may be wrong.
+- `/health` distinguishes `STARTING`, `READY`, and `FAILED`.
+- `/api/v1/identity` exposes the immutable artifact, runtime limits, source release
+  marker, and receipt boundary. Source checksums establish internal bundle
+  consistency only; they are not external authorship evidence.
+- Prompts are not intentionally persisted by this source.
+
+The upstream training and evaluation receipts are checked against the repository's
+declared Ed25519 owner key and chained canonical payload hash. That is **owner-key
+continuity**, not independent authorship evidence. Those receipts do not cover the
+GGUF quantization, this Space's source, runtime outputs, independent benchmarking,
+or safety certification.
+
+## Attribution and licenses
+
+Space source: Apache-2.0, copyright SZL HOLDINGS LLC.
+
+- Runtime model: [SZL-Khipu-1.5B-GGUF](https://huggingface.co/SZLHOLDINGS/SZL-Khipu-1.5B-GGUF), Apache-2.0.
+- Fine-tuned model: [SZL-Khipu-1.5B-BrainNavigator](https://huggingface.co/SZLHOLDINGS/SZL-Khipu-1.5B-BrainNavigator), Apache-2.0.
+- Base model: [Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct), Apache-2.0.
+- Inference binding: [llama-cpp-python v0.3.21](https://github.com/abetlen/llama-cpp-python/releases/tag/v0.3.21), MIT; CPU wheel URL and SHA-256 are pinned in `requirements.txt`.
+
+No independent benchmark, post-quantization evaluation, or safety certification is claimed.

app.py

@@ -0,0 +1,366 @@
+from __future__ import annotations
+
+import base64
+import hashlib
+import json
+import os
+import platform
+import threading
+import time
+import urllib.request
+from contextlib import asynccontextmanager
+from pathlib import Path
+from typing import Any
+
+from fastapi import FastAPI, HTTPException, Request
+from fastapi.responses import HTMLResponse, JSONResponse
+from huggingface_hub import hf_hub_download
+from pydantic import BaseModel, Field, field_validator
+
+
+SPACE_ID = "SZLHOLDINGS/szl-model-inference-lab"
+MODEL_REPO = "SZLHOLDINGS/SZL-Khipu-1.5B-GGUF"
+MODEL_REVISION = "67d60ec577730747055491640cfb91fc4a4b5d25"
+MODEL_FILE = "SZL-Khipu-1.5B-Q4_K_M.gguf"
+MODEL_SIZE = 986_047_904
+MODEL_SHA256 = "13c1a1993063e1dff92f7413ccf48eaca6d48efc8801ae9af35961ae3396623a"
+RECEIPT_FILES = (
+    "training_receipt.signed.json",
+    "eval_receipt.signed.json",
+    "owner_pubkey.json",
+)
+MAX_INPUT_CHARS = 1_200
+MAX_NEW_TOKENS = 32
+INFERENCE_BUDGET_SECONDS = 45.0
+SOURCE_ROOT = Path(__file__).resolve().parent
+
+
+state: dict[str, Any] = {
+    "status": "STARTING",
+    "failure_code": None,
+    "model_path": None,
+    "model_sha256": None,
+    "source_integrity": False,
+    "receipt_status": "NOT_CHECKED",
+    "llama_cpp_version": None,
+}
+llm: Any = None
+inference_lock = threading.Lock()
+
+
+def sha256_file(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as handle:
+        for chunk in iter(lambda: handle.read(8 * 1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def load_release_manifest() -> dict[str, Any]:
+    manifest = json.loads((SOURCE_ROOT / "release.json").read_text(encoding="utf-8"))
+    for relative, expected in manifest["source_files"].items():
+        if sha256_file(SOURCE_ROOT / relative) != expected:
+            raise RuntimeError("SOURCE_INTEGRITY_MISMATCH")
+    return manifest
+
+
+def artifact_path(filename: str) -> Path:
+    override = os.getenv("MODEL_DIR_OVERRIDE")
+    if override:
+        return Path(override) / filename
+    return Path(
+        hf_hub_download(
+            repo_id=MODEL_REPO,
+            filename=filename,
+            revision=MODEL_REVISION,
+            local_files_only=True,
+            token=False,
+        )
+    )
+
+
+def verify_receipts() -> dict[str, Any]:
+    from cryptography.hazmat.primitives.serialization import load_der_public_key
+
+    owner = json.loads(artifact_path("owner_pubkey.json").read_text(encoding="utf-8"))
+    receipts: dict[str, dict[str, Any]] = {}
+    for filename in RECEIPT_FILES[:2]:
+        receipt = json.loads(artifact_path(filename).read_text(encoding="utf-8"))
+        canonical = json.dumps(
+            receipt["payload"], ensure_ascii=False, separators=(",", ":"), sort_keys=True
+        )
+        if canonical != receipt["canonical"]:
+            raise RuntimeError("RECEIPT_CANONICAL_MISMATCH")
+        if receipt["publicKeySpkiBase64"] != owner["publicKeySpkiBase64"]:
+            raise RuntimeError("RECEIPT_KEY_MISMATCH")
+        public_key = load_der_public_key(base64.b64decode(receipt["publicKeySpkiBase64"]))
+        public_key.verify(
+            base64.b64decode(receipt["signatureBase64"]), canonical.encode("utf-8")
+        )
+        receipts[filename] = receipt
+    training_digest = hashlib.sha256(
+        receipts["training_receipt.signed.json"]["canonical"].encode("utf-8")
+    ).hexdigest()
+    if (
+        receipts["eval_receipt.signed.json"]["payload"]["trainingReceiptSha256"]
+        != training_digest
+    ):
+        raise RuntimeError("RECEIPT_CHAIN_MISMATCH")
+    return {
+        "status": "OWNER_SIGNATURES_VALID",
+        "key_id": owner["keyId"],
+        "training_canonical_sha256": training_digest,
+        "eval_canonical_sha256": hashlib.sha256(
+            receipts["eval_receipt.signed.json"]["canonical"].encode("utf-8")
+        ).hexdigest(),
+    }
+
+
+def initialize() -> None:
+    global llm
+    try:
+        manifest = load_release_manifest()
+        state["source_integrity"] = True
+        state["release_id"] = manifest["release_id"]
+
+        model_path = artifact_path(MODEL_FILE)
+        if model_path.stat().st_size != MODEL_SIZE:
+            raise RuntimeError("MODEL_SIZE_MISMATCH")
+        actual_sha = sha256_file(model_path)
+        if actual_sha != MODEL_SHA256:
+            raise RuntimeError("MODEL_SHA256_MISMATCH")
+        state["model_path"] = str(model_path)
+        state["model_sha256"] = actual_sha
+
+        receipt = verify_receipts()
+        state["receipt_status"] = receipt["status"]
+        state["receipt_evidence"] = receipt
+
+        import llama_cpp
+        from llama_cpp import Llama
+
+        state["llama_cpp_version"] = llama_cpp.__version__
+        threads = max(1, min(int(os.getenv("CPU_CORES", "2")), 2))
+        llm = Llama(
+            model_path=str(model_path),
+            n_ctx=1024,
+            n_batch=64,
+            n_threads=threads,
+            n_threads_batch=threads,
+            seed=0,
+            use_mmap=True,
+            use_mlock=False,
+            verbose=False,
+        )
+        state["status"] = "READY"
+    except Exception as exc:  # keep /health available for honest diagnostics
+        state["status"] = "FAILED"
+        state["failure_code"] = str(exc) if str(exc).isupper() else type(exc).__name__
+
+
+@asynccontextmanager
+async def lifespan(_: FastAPI):
+    state["status"] = "STARTING"
+    initialize()
+    yield
+
+
+app = FastAPI(
+    title="SZL Model Inference Lab",
+    version="1.0.0",
+    docs_url=None,
+    redoc_url=None,
+    openapi_url="/api/openapi.json",
+    lifespan=lifespan,
+)
+
+
+@app.middleware("http")
+async def bounded_requests(request: Request, call_next):
+    if request.method == "POST":
+        raw_length = request.headers.get("content-length")
+        if raw_length:
+            try:
+                content_length = int(raw_length)
+            except ValueError:
+                return JSONResponse({"detail": "invalid content-length"}, status_code=400)
+            if content_length < 0:
+                return JSONResponse({"detail": "invalid content-length"}, status_code=400)
+            if content_length > 8192:
+                return JSONResponse({"detail": "request body too large"}, status_code=413)
+    response = await call_next(request)
+    response.headers["X-Content-Type-Options"] = "nosniff"
+    response.headers["Referrer-Policy"] = "no-referrer"
+    response.headers["X-Frame-Options"] = "SAMEORIGIN"
+    return response
+
+
+class InferenceRequest(BaseModel):
+    prompt: str = Field(min_length=1, max_length=MAX_INPUT_CHARS)
+    max_new_tokens: int = Field(default=24, ge=1, le=MAX_NEW_TOKENS)
+
+    @field_validator("prompt")
+    @classmethod
+    def clean_prompt(cls, value: str) -> str:
+        value = value.strip()
+        if not value or "\x00" in value:
+            raise ValueError("prompt must contain visible text and no NUL bytes")
+        return value
+
+
+class TimeBudgetStop:
+    def __init__(self, started: float, budget: float) -> None:
+        self.started = started
+        self.budget = budget
+
+    def __call__(self, _input_ids: Any, _logits: Any) -> bool:
+        return time.monotonic() - self.started >= self.budget
+
+
+def identity_payload() -> dict[str, Any]:
+    manifest = json.loads((SOURCE_ROOT / "release.json").read_text(encoding="utf-8"))
+    return {
+        "schema": "szl.hf-free-inference-identity/v1",
+        "status": state["status"],
+        "space": {
+            "id": os.getenv("SPACE_ID", SPACE_ID),
+            "release_id": manifest["release_id"],
+            "source_integrity": state["source_integrity"],
+            "source_integrity_meaning": (
+                "internal release-file checksum consistency; not external authorship evidence"
+            ),
+            "license": "Apache-2.0",
+        },
+        "hardware": {
+            "required": "cpu-basic",
+            "accelerator_observed": os.getenv("ACCELERATOR", "none"),
+            "cpu_cores_observed": os.getenv("CPU_CORES", "unknown"),
+            "memory_observed": os.getenv("MEMORY", "unknown"),
+        },
+        "model": {
+            "repo": MODEL_REPO,
+            "revision": MODEL_REVISION,
+            "file": MODEL_FILE,
+            "size": MODEL_SIZE,
+            "sha256_expected": MODEL_SHA256,
+            "sha256_loaded": state["model_sha256"],
+            "base_model": "Qwen/Qwen2.5-1.5B-Instruct",
+            "license": "Apache-2.0",
+        },
+        "runtime": {
+            "python": platform.python_version(),
+            "llama_cpp_python": state["llama_cpp_version"],
+            "concurrency": 1,
+            "max_input_chars": MAX_INPUT_CHARS,
+            "max_new_tokens": MAX_NEW_TOKENS,
+            "inference_budget_seconds": INFERENCE_BUDGET_SECONDS,
+        },
+        "receipt_boundary": {
+            "status": state["receipt_status"],
+            "evidence": state.get("receipt_evidence"),
+            "covers": "owner-key continuity for upstream training/eval receipt payloads",
+            "does_not_cover": [
+                "Space source authorship",
+                "GGUF quantization quality",
+                "this runtime's outputs",
+                "independent benchmarking",
+                "safety certification",
+            ],
+        },
+        "failure_code": state["failure_code"],
+    }
+
+
+@app.get("/health")
+def health() -> JSONResponse:
+    code = 200 if state["status"] == "READY" else 503
+    return JSONResponse(
+        {
+            "status": state["status"],
+            "model_sha256_verified": state["model_sha256"] == MODEL_SHA256,
+            "source_integrity": state["source_integrity"],
+            "receipt_status": state["receipt_status"],
+            "failure_code": state["failure_code"],
+        },
+        status_code=code,
+    )
+
+
+@app.get("/api/v1/identity")
+def identity() -> dict[str, Any]:
+    return identity_payload()
+
+
+@app.post("/api/v1/infer")
+def infer(request: InferenceRequest) -> dict[str, Any]:
+    if state["status"] != "READY" or llm is None:
+        raise HTTPException(status_code=503, detail="runtime is not ready")
+    if not inference_lock.acquire(blocking=False):
+        raise HTTPException(status_code=429, detail="one inference is already running")
+    started = time.monotonic()
+    try:
+        stream = llm.create_chat_completion(
+            messages=[
+                {
+                    "role": "system",
+                    "content": (
+                        "You are a bounded research demo. Answer briefly. If evidence is missing, "
+                        "say so; do not claim independent benchmarking or safety certification."
+                    ),
+                },
+                {"role": "user", "content": request.prompt},
+            ],
+            max_tokens=request.max_new_tokens,
+            temperature=0.0,
+            top_p=1.0,
+            top_k=1,
+            stop=["<|im_end|>", "<|endoftext|>"],
+            stream=True,
+        )
+        chunks: list[str] = []
+        finish_reason = None
+        timed_out = False
+        try:
+            for event in stream:
+                if time.monotonic() - started >= INFERENCE_BUDGET_SECONDS:
+                    timed_out = True
+                    break
+                choice = event["choices"][0]
+                chunks.append(choice.get("delta", {}).get("content") or "")
+                finish_reason = choice.get("finish_reason") or finish_reason
+        finally:
+            close = getattr(stream, "close", None)
+            if close:
+                close()
+        output = "".join(chunks).strip()
+        return {
+            "output": output,
+            "elapsed_ms": round((time.monotonic() - started) * 1000),
+            "finish_reason": "time_budget" if timed_out else finish_reason,
+            "model": {
+                "repo": MODEL_REPO,
+                "revision": MODEL_REVISION,
+                "file": MODEL_FILE,
+                "sha256": MODEL_SHA256,
+            },
+            "determinism": "greedy temperature=0; exact bytes and runtime are disclosed",
+        }
+    finally:
+        inference_lock.release()
+
+
+@app.get("/", response_class=HTMLResponse)
+def index() -> str:
+    return """<!doctype html>
+<html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1">
+<title>SZL Model Inference Lab</title><style>
+body{margin:0;background:#080b13;color:#e9efff;font:16px/1.5 system-ui,sans-serif}main{max-width:820px;margin:auto;padding:40px 22px}
+.card{background:#101728;border:1px solid #2a385b;border-radius:18px;padding:24px;box-shadow:0 18px 60px #0007}h1{margin:.1em 0;color:#8ce8ff}
+textarea{box-sizing:border-box;width:100%;min-height:130px;background:#080d19;color:#fff;border:1px solid #40537f;border-radius:10px;padding:12px}
+button{margin-top:12px;background:#7ce6ff;color:#06101a;border:0;border-radius:999px;padding:10px 18px;font-weight:700;cursor:pointer}pre{white-space:pre-wrap;background:#080d19;padding:14px;border-radius:10px;min-height:52px}code{color:#a7f3d0}.fine{color:#aebbd4;font-size:.9rem}a{color:#8ce8ff}</style></head>
+<body><main><div class="card"><p class="fine">FREE CPU BASIC · ONE REQUEST AT A TIME · IMMUTABLE Q4_K_M</p><h1>SZL Model Inference Lab</h1>
+<p>Real, bounded CPU inference for <code>SZL-Khipu-1.5B-GGUF</code>. Max 1,200 input characters and 32 generated tokens.</p>
+<textarea id="prompt" maxlength="1200">Reply with one short sentence describing what a cryptographic receipt can prove.</textarea><br>
+<button id="run">Run bounded inference</button><pre id="out">Ready.</pre>
+<p class="fine">Owner-signed upstream receipts prove owner-key continuity only. No independent benchmark, post-quantization quality claim, or safety certification. Prompts are not intentionally stored. <a href="/api/v1/identity">Machine identity</a> · <a href="/health">Health</a></p></div></main>
+<script>const b=document.querySelector('#run'),o=document.querySelector('#out'),p=document.querySelector('#prompt');b.onclick=async()=>{b.disabled=true;o.textContent='Running on free CPU…';try{const r=await fetch('/api/v1/infer',{method:'POST',headers:{'content-type':'application/json'},body:JSON.stringify({prompt:p.value,max_new_tokens:24})});const j=await r.json();o.textContent=r.ok?j.output:JSON.stringify(j)}catch(e){o.textContent='Request failed: '+e}finally{b.disabled=false}}</script></body></html>"""

release.json

@@ -0,0 +1,13 @@
+{
+  "schema": "szl.space-source-release/v1",
+  "release_id": "3b817b07-873a-4c11-ada3-72364fcf6731",
+  "source_files": {
+    ".dockerignore": "2f498f75a9cf48b332ab65dbdd1d387279c11eae9cdf7131e406f7cae61f709a",
+    "Dockerfile": "eb022050a39b0eafaa8916a883e94ca35f378308503262d6c3f1e239eda6444e",
+    "LICENSE": "0a41cda63c0751e5cdb240864525fc8bf29b2d3f2cb5c6406b9bae97f1e1407b",
+    "README.md": "9f46d43f4517fd44d85257a1abfcf5be5ba7387aa4a37dfdd1f7aeccc7685f4a",
+    "app.py": "3ed2bd3302e8dcd994ae4530bea827f9592b1d3af0b2ce69f76be8f490036004",
+    "requirements.txt": "74e8d7b61052bff09996927411d40603da8534e3953726303073fc6f6ec778f4",
+    "tests/test_app.py": "8a85490aee58bb6ad1368898d68ffc85d55279230c9c42cf9fdee7dbed716c81"
+  }
+}

requirements.txt

@@ -0,0 +1,6 @@
+fastapi==0.139.0
+uvicorn==0.51.0
+huggingface-hub==1.23.0
+pydantic==2.13.4
+cryptography==49.0.0
+llama-cpp-python @ https://github.com/abetlen/llama-cpp-python/releases/download/v0.3.21/llama_cpp_python-0.3.21-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=d9d4e0cc1a7f779dd224d0e07ac8ea793a58b345723ba90cd68ee20acf85ad68

tests/test_app.py

@@ -0,0 +1,50 @@
+import json
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+
+import app
+
+
+class AppContractTests(unittest.TestCase):
+    def test_immutable_model_contract(self):
+        self.assertEqual(len(app.MODEL_REVISION), 40)
+        self.assertEqual(len(app.MODEL_SHA256), 64)
+        self.assertEqual(app.MODEL_SIZE, 986_047_904)
+        self.assertEqual(app.MAX_NEW_TOKENS, 32)
+
+    def test_prompt_contract(self):
+        self.assertEqual(app.InferenceRequest(prompt="  hello  ").prompt, "hello")
+        with self.assertRaises(ValueError):
+            app.InferenceRequest(prompt=" ")
+        with self.assertRaises(ValueError):
+            app.InferenceRequest(prompt="x" * 1_201)
+
+    def test_time_budget_stop(self):
+        with patch("app.time.monotonic", return_value=10.0):
+            stop = app.TimeBudgetStop(0.0, 5.0)
+            self.assertTrue(stop(None, None))
+
+    def test_generation_budget_is_bounded(self):
+        self.assertLessEqual(app.INFERENCE_BUDGET_SECONDS, 45.0)
+        self.assertLessEqual(app.MAX_NEW_TOKENS, 32)
+
+    def test_sha256_file(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / "sample"
+            path.write_bytes(b"abc")
+            self.assertEqual(
+                app.sha256_file(path),
+                "ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad",
+            )
+
+    def test_identity_is_explicit_about_receipt_boundary(self):
+        payload = app.identity_payload()
+        boundary = json.dumps(payload["receipt_boundary"])
+        self.assertIn("independent benchmarking", boundary)
+        self.assertIn("safety certification", boundary)
+
+
+if __name__ == "__main__":
+    unittest.main()