qubrid 0.1.0__tar.gz

qubrid-0.1.0/.github/workflows/test.yml

@@ -0,0 +1,26 @@
+name: tests
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.9", "3.11", "3.12"]
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install
+        run: pip install -e ".[dev]"
+      - name: Lint
+        run: ruff check src tests examples
+      - name: Format check
+        run: ruff format --check src tests examples
+      - name: Tests
+        run: python -m pytest tests/ -q

qubrid-0.1.0/.gitignore

@@ -0,0 +1,9 @@
+__pycache__/
+*.egg-info/
+dist/
+build/
+.venv/
+.pytest_cache/
+.ruff_cache/
+qubrid-docs/
+qubrid-examples/

qubrid-0.1.0/AGENTS.md

@@ -0,0 +1,27 @@
+# AGENTS.md
+
+First-party Python SDK for Qubrid AI inference (`qubrid` on PyPI). Standalone: `httpx` + `pydantic` only — never add an `openai` dependency.
+
+## Commands
+
+- Install (editable + dev): `pip install -e ".[dev]"`
+- Tests: `python -m pytest tests/ -q` (27 tests, `respx`-mocked HTTP — no network/key needed)
+- Lint: `ruff check src tests` · Format: `ruff format src tests`
+- `Optional[...]` is deliberate (py39 runtime compat for pydantic); `UP006/7/35/45` are silenced in `pyproject.toml` — do not "modernize" to `X | None`.
+
+## Architecture
+
+- `src/qubrid/client.py` — `QubridClient` / `AsyncQubridClient` (env `QUBRID_API_KEY`, `base_url` + `legacy_base_url`, `timeout`, `max_retries`). Owns one `httpx` client; resources are thin wrappers over it.
+- `src/qubrid/_http.py` — retry-with-backoff on 429/5xx (honors `Retry-After`), error mapping, SSE `data:` parsing. Sync/async parity lives here.
+- `src/qubrid/_types.py` — pydantic models mirroring `QubridAI-Inc/docs` `api-reference/*/openapi.json`. Model IDs are plain `str` (catalog churns).
+- Resources: `chat.py` (`POST /chat/completions`, streaming + not), `vision_ocr.py` (same path, docs-example defaults), `audio.py` (TTS on `/v1`, STT multipart auto-routed to legacy `/api/v1/qubridai`), `extra.py` (embeddings/images/video).
+- `src/qubrid/exceptions.py` — `QubridError` + typed subclasses carrying `status_code/code/type/param/request_id/body`.
+
+## Gotchas (verified vs docs repo + examples repo, 2026-09-08)
+
+- Two live prefixes: canonical `https://platform.qubrid.com/v1`, legacy `.../api/v1/qubridai` for STT **only**. Do not mix field names across them.
+- Embeddings has **no** Qubrid OpenAPI spec — `embeddings.create` is an explicitly experimental OpenAI-compatible passthrough (user-approved). Flag shape changes.
+- Image-edits has no OpenAPI file (quickstarts only): local files → multipart, remote URLs → JSON `images:[...]`. Video accepts both `url` and `video_url` (`resolved_url`).
+- No REST API exists for RAG / fine-tuning / GPU provisioning / models listing (UI-only) — do not invent endpoints; raise/flag instead.
+- `api-reference/endpoint/` (`/plants`) is a Mintlify template placeholder, not real.
+- OCR default `tencent/HunyuanOCR` (live-verified 2026-09-08; examples' `Hunyuan/Hunyuan-OCR-1B` 404s `model_not_found`).

qubrid-0.1.0/CHANGELOG.md

@@ -0,0 +1,26 @@
+# Changelog
+
+All notable changes to the `qubrid` package. Format follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
+
+## [0.1.0] — 2026-09-08
+
+### Added
+- Initial standalone SDK (`httpx` + `pydantic`, no `openai` dependency).
+- `QubridClient` / `AsyncQubridClient` with `QUBRID_API_KEY` env fallback, `base_url` + `legacy_base_url` overrides, configurable timeout/retries.
+- `chat.completions.create` (streaming SSE generator + non-streaming), live-verified vs `api-reference/chat/openapi.json`.
+- `vision.analyze`, `ocr.extract` thin wrappers over `/chat/completions`, live-verified (OCR default corrected to `tencent/HunyuanOCR` — examples' `Hunyuan/Hunyuan-OCR-1B` 404s).
+- `audio.transcriptions.create` (STT multipart, auto-routed to legacy `/api/v1/qubridai/audio/transcribe`), live-verified.
+- `audio.speech.create` (TTS `POST /audio/generations`), `images.generate` / `images.edit`, `videos.generate` (accepts both `url` and `video_url`), `embeddings.create` — retained per-spec but **not working live** (see Fixed/Unverified below).
+- Typed errors (`AuthenticationError`, `RateLimitError`, `InvalidRequestError`, … with `code`/`request_id`/`body`), retry-with-backoff on 429/5xx.
+- Mocked test suite (`respx`), README quickstart + per-category examples.
+
+### Live-verification findings (2026-09-08 smoke test, ~$0.0005 spend)
+- Fixed: OCR default `Hunyuan/Hunyuan-OCR-1B` → `tencent/HunyuanOCR` (former is 404 `model_not_found` live; latter verified 200 with usage 1699/17).
+- Unverified: TTS returns `400 content_policy_violation` on all tested inputs including the docs' own example sentence; `images/generations` returns `405 Method Not Allowed` despite exact spec match (`images.edit`, `videos.generate` deliberately untested, same family); `POST /v1/embeddings` is 404 with empty body (endpoint doesn't exist).
+
+### Final verification (2026-09-08, v0.1.0 polish)
+- Live-verified end-to-end against the real Qubrid API: chat (streaming + non-streaming), vision, OCR (corrected default `tencent/HunyuanOCR`), STT.
+- Confirmed non-functional: embeddings (endpoint doesn't exist, 404 with empty body); images/video generation (`405 Method Not Allowed` despite exact OpenAPI spec match — likely account/region routing issue on Qubrid's side; siblings untested by decision); TTS (reachable, request shape accepted, but safety filter rejected all tested inputs including Qubrid's own documented example sentence).
+- Clean-room install verified: fresh venv, non-editable `pip install`, import + live quickstart call succeeded from outside the source tree.
+- Total live-testing spend across the full verification effort: ~$0.0006.
+- Docs-only fixes: streaming example guards against `None` deltas; README notes Windows console unicode quirk (`PYTHONIOENCODING=utf-8`). No core SDK logic changed.

qubrid-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Qubrid AI Community SDK
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

qubrid-0.1.0/PKG-INFO

@@ -0,0 +1,190 @@
+Metadata-Version: 2.5
+Name: qubrid
+Version: 0.1.0
+Summary: First-party Python SDK for Qubrid AI GPU-accelerated inference (chat, vision, OCR, TTS, STT, images, video, embeddings)
+Project-URL: Homepage, https://platform.qubrid.com
+Project-URL: Documentation, https://github.com/QubridAI-Inc/docs
+Project-URL: Repository, https://github.com/QubridAI-Inc/qubrid-examples
+Project-URL: Changelog, https://github.com/Abhi-shek26/python-sdk-buildup/blob/main/CHANGELOG.md
+Author-email: Qubrid AI Community SDK <support@qubrid.com>
+License: MIT
+License-File: LICENSE
+Keywords: ai,chat,inference,llm,qubrid,sdk,stt,tts,vision
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.9
+Requires-Dist: httpx<1,>=0.27
+Requires-Dist: pydantic>=2.0
+Provides-Extra: dev
+Requires-Dist: mypy>=1.0; extra == 'dev'
+Requires-Dist: pytest-asyncio>=0.23; extra == 'dev'
+Requires-Dist: pytest>=8; extra == 'dev'
+Requires-Dist: respx>=0.22; extra == 'dev'
+Requires-Dist: ruff>=0.4; extra == 'dev'
+Description-Content-Type: text/markdown
+
+# Qubrid Python SDK — `qubrid`
+
+First-party, standalone Python client for [Qubrid AI](https://platform.qubrid.com) GPU-accelerated inference. No `openai` dependency (`httpx` + `pydantic` only).
+
+**Docs truth:** endpoint shapes verified against [`QubridAI-Inc/docs`](https://github.com/QubridAI-Inc/docs) (`api-reference/*/openapi.json`) and runnable payloads in [`QubridAI-Inc/qubrid-examples`](https://github.com/QubridAI-Inc/qubrid-examples).
+
+| Resource | Endpoint | Base |
+|---|---|---|
+| `client.chat.completions` | `POST /chat/completions` (OpenAI-compatible, `stream` supported) | `…/v1` |
+| `client.vision` | same `POST /chat/completions` (image_url parts) | `…/v1` |
+| `client.ocr` | same `POST /chat/completions` (default `tencent/HunyuanOCR`, live-verified) | `…/v1` |
+| `client.audio.speech` (TTS) | `POST /audio/generations` `{model,text,voice,language_type}` | `…/v1` |
+| `client.audio.transcriptions` (STT) | `POST /audio/transcribe` multipart `{file,model}` | `…/api/v1/qubridai` (legacy — auto-routed) |
+| `client.images` | `POST /images/generations`, `POST /images/edits` | `…/v1` |
+| `client.videos` | `POST /videos/generations` | `…/v1` |
+| `client.embeddings` | `POST /embeddings` — ⚠️ **EXPERIMENTAL**, no Qubrid OpenAPI spec | `…/v1` |
+
+**Not in the SDK (no documented REST API — UI-only):** RAG (beta upload + Chat UI), no-code Fine-Tuning studio, GPU Instances/Clusters/Bare-Metal provisioning, models listing (`GET /models`). Model IDs are plain `str` (catalog of 100+ changes often — never an enum).
+
+## Live verification status (smoke-tested 2026-09-08)
+
+Confirmed working end-to-end: chat (+ streaming), vision, OCR (default `tencent/HunyuanOCR`), STT.
+
+- **TTS — unverified:** endpoint reachable, shape accepted, but the safety filter rejected every tested input (`400 content_policy_violation`), including the docs' own example sentence.
+- **Images / video — currently unroutable:** `POST /v1/images/generations` returns `405 Method Not Allowed` despite matching the OpenAPI spec exactly; `images.edit` / `videos.generate` untested (same family, deliberately not probed). Code retained per-spec; treat as unstable.
+- **Embeddings — doesn't exist:** `POST /v1/embeddings` returns 404 with an empty body. The method is kept as an experimental stub only.
+
+## Install
+
+```bash
+pip install qubrid
+# or from source:
+pip install -e ".[dev]"
+```
+
+Requires Python 3.9+, `QUBRID_API_KEY` from https://platform.qubrid.com/api-keys:
+
+```bash
+export QUBRID_API_KEY="qb_..."
+```
+
+## Quickstart
+
+```python
+from qubrid import QubridClient
+
+client = QubridClient()  # reads QUBRID_API_KEY
+
+resp = client.chat.completions.create(
+    model="openai/gpt-oss-120b",
+    messages=[{"role": "user", "content": "Summarize this ticket into next steps."}],
+    max_tokens=4096,
+    temperature=0.7,
+)
+print(resp.content)
+print(resp.usage.total_tokens)
+
+# Streaming
+for chunk in client.chat.completions.create(
+    model="deepseek-ai/DeepSeek-V3.2",
+    messages=[{"role": "user", "content": "Write a haiku about GPUs."}],
+    stream=True,
+):
+    print(chunk.delta, end="", flush=True)
+
+# Async
+import asyncio
+from qubrid import AsyncQubridClient
+
+async def main():
+    async with AsyncQubridClient() as c:
+        r = await c.chat.completions.create(
+            model="openai/gpt-oss-120b", messages=[{"role": "user", "content": "Hi"}])
+        print(r.content)
+
+asyncio.run(main())
+```
+
+## One example per category
+
+```python
+# Vision (Qwen/Qwen3-VL-Plus + image_url parts)
+v = client.vision.analyze(
+    prompt="What is in this image?",
+    image_urls="https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg",
+)
+print(v.content)
+
+# OCR (default tencent/HunyuanOCR — live-verified; examples' Hunyuan/Hunyuan-OCR-1B 404s)
+o = client.ocr.extract(image_urls="https://example.com/receipt.jpg")
+print(o.content)
+
+# TTS → hosted mp3 URL
+s = client.audio.speech.create(text="Today is a wonderful day to build something people love!",
+                               voice="Cherry", language_type="Auto")
+print(s.url)
+
+# STT (multipart upload, auto-routed to legacy prefix)
+t = client.audio.transcriptions.create(model="openai/whisper-large-v3", file="clip.wav")
+print(t.text)
+
+# Images
+img = client.images.generate(model="p-image", prompt="a lighthouse at dusk",
+                             aspect_ratio="16:9", output_format="webp")
+print(img.data[0].url)
+edit = client.images.edit(model="p-image-edit", prompt="add a red flag",
+                          image="https://example.com/in.png")
+
+# Video (only documented model: p-video, ≤10s)
+vid = client.videos.generate(model="p-video", prompt="ocean waves", duration=5)
+print(vid.url)
+
+# Embeddings — EXPERIMENTAL passthrough (no Qubrid spec; may change)
+e = client.embeddings.create(model="BAAI/bge-large-en-v1.5", input="hello world")
+print(e.data[0].embedding[:4])
+```
+
+More in [`examples/`](examples/).
+
+## Configuration
+
+```python
+QubridClient(
+    api_key="qb_...",              # or QUBRID_API_KEY env
+    base_url="https://platform.qubrid.com/v1",                 # override canonical
+    legacy_base_url="https://platform.qubrid.com/api/v1/qubridai",  # STT only
+    timeout=60.0,                  # seconds; per-call `timeout=` also supported
+    max_retries=2,                 # retry-with-backoff on 429/5xx
+)
+```
+
+## Errors
+
+```python
+from qubrid import QubridError, AuthenticationError, RateLimitError
+
+try:
+    client.chat.completions.create(model="m", messages=[])
+except AuthenticationError:
+    ...  # bad/missing key (401)
+except RateLimitError as e:
+    print(e.status_code, e.code, e.request_id, e.body)
+except QubridError as e:
+    ...  # base class: InvalidRequestError, NotFoundError,
+         # PermissionDeniedError, QuotaExceededError, ServerError, APIConnectionError
+```
+
+## Docs & flags
+
+- STT lives on the legacy prefix — the client handles it; only override `legacy_base_url` for staging/mocks.
+- Live-verified 2026-09-08: runnable examples' `Hunyuan/Hunyuan-OCR-1B` returns 404 `model_not_found`; the working id is `tencent/HunyuanOCR` (the SDK default).
+- Video quickstart JS reads `data[0].video_url` while the spec says `url` — both accepted (`VideoData.resolved_url`).
+- Image-edit has no OpenAPI file (quickstarts only); local files → multipart, remote URLs → JSON `images:[...]`.
+- Anything under `api-reference/endpoint/` (`/plants`) is a Mintlify template placeholder, not a real API.
+
+## Known environment quirks
+
+- **Garbled unicode on Windows terminals:** model output containing emoji, accented characters, or other non-ASCII text may render as `�` in default Windows consoles (e.g. `cmd.exe` / PowerShell with a legacy code page). This is a console-encoding artifact, not an SDK bug — the returned strings are correct. If affected, run with `PYTHONIOENCODING=utf-8` or use a UTF-8-capable terminal (e.g. Windows Terminal).

qubrid-0.1.0/README.md

@@ -0,0 +1,158 @@
+# Qubrid Python SDK — `qubrid`
+
+First-party, standalone Python client for [Qubrid AI](https://platform.qubrid.com) GPU-accelerated inference. No `openai` dependency (`httpx` + `pydantic` only).
+
+**Docs truth:** endpoint shapes verified against [`QubridAI-Inc/docs`](https://github.com/QubridAI-Inc/docs) (`api-reference/*/openapi.json`) and runnable payloads in [`QubridAI-Inc/qubrid-examples`](https://github.com/QubridAI-Inc/qubrid-examples).
+
+| Resource | Endpoint | Base |
+|---|---|---|
+| `client.chat.completions` | `POST /chat/completions` (OpenAI-compatible, `stream` supported) | `…/v1` |
+| `client.vision` | same `POST /chat/completions` (image_url parts) | `…/v1` |
+| `client.ocr` | same `POST /chat/completions` (default `tencent/HunyuanOCR`, live-verified) | `…/v1` |
+| `client.audio.speech` (TTS) | `POST /audio/generations` `{model,text,voice,language_type}` | `…/v1` |
+| `client.audio.transcriptions` (STT) | `POST /audio/transcribe` multipart `{file,model}` | `…/api/v1/qubridai` (legacy — auto-routed) |
+| `client.images` | `POST /images/generations`, `POST /images/edits` | `…/v1` |
+| `client.videos` | `POST /videos/generations` | `…/v1` |
+| `client.embeddings` | `POST /embeddings` — ⚠️ **EXPERIMENTAL**, no Qubrid OpenAPI spec | `…/v1` |
+
+**Not in the SDK (no documented REST API — UI-only):** RAG (beta upload + Chat UI), no-code Fine-Tuning studio, GPU Instances/Clusters/Bare-Metal provisioning, models listing (`GET /models`). Model IDs are plain `str` (catalog of 100+ changes often — never an enum).
+
+## Live verification status (smoke-tested 2026-09-08)
+
+Confirmed working end-to-end: chat (+ streaming), vision, OCR (default `tencent/HunyuanOCR`), STT.
+
+- **TTS — unverified:** endpoint reachable, shape accepted, but the safety filter rejected every tested input (`400 content_policy_violation`), including the docs' own example sentence.
+- **Images / video — currently unroutable:** `POST /v1/images/generations` returns `405 Method Not Allowed` despite matching the OpenAPI spec exactly; `images.edit` / `videos.generate` untested (same family, deliberately not probed). Code retained per-spec; treat as unstable.
+- **Embeddings — doesn't exist:** `POST /v1/embeddings` returns 404 with an empty body. The method is kept as an experimental stub only.
+
+## Install
+
+```bash
+pip install qubrid
+# or from source:
+pip install -e ".[dev]"
+```
+
+Requires Python 3.9+, `QUBRID_API_KEY` from https://platform.qubrid.com/api-keys:
+
+```bash
+export QUBRID_API_KEY="qb_..."
+```
+
+## Quickstart
+
+```python
+from qubrid import QubridClient
+
+client = QubridClient()  # reads QUBRID_API_KEY
+
+resp = client.chat.completions.create(
+    model="openai/gpt-oss-120b",
+    messages=[{"role": "user", "content": "Summarize this ticket into next steps."}],
+    max_tokens=4096,
+    temperature=0.7,
+)
+print(resp.content)
+print(resp.usage.total_tokens)
+
+# Streaming
+for chunk in client.chat.completions.create(
+    model="deepseek-ai/DeepSeek-V3.2",
+    messages=[{"role": "user", "content": "Write a haiku about GPUs."}],
+    stream=True,
+):
+    print(chunk.delta, end="", flush=True)
+
+# Async
+import asyncio
+from qubrid import AsyncQubridClient
+
+async def main():
+    async with AsyncQubridClient() as c:
+        r = await c.chat.completions.create(
+            model="openai/gpt-oss-120b", messages=[{"role": "user", "content": "Hi"}])
+        print(r.content)
+
+asyncio.run(main())
+```
+
+## One example per category
+
+```python
+# Vision (Qwen/Qwen3-VL-Plus + image_url parts)
+v = client.vision.analyze(
+    prompt="What is in this image?",
+    image_urls="https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg",
+)
+print(v.content)
+
+# OCR (default tencent/HunyuanOCR — live-verified; examples' Hunyuan/Hunyuan-OCR-1B 404s)
+o = client.ocr.extract(image_urls="https://example.com/receipt.jpg")
+print(o.content)
+
+# TTS → hosted mp3 URL
+s = client.audio.speech.create(text="Today is a wonderful day to build something people love!",
+                               voice="Cherry", language_type="Auto")
+print(s.url)
+
+# STT (multipart upload, auto-routed to legacy prefix)
+t = client.audio.transcriptions.create(model="openai/whisper-large-v3", file="clip.wav")
+print(t.text)
+
+# Images
+img = client.images.generate(model="p-image", prompt="a lighthouse at dusk",
+                             aspect_ratio="16:9", output_format="webp")
+print(img.data[0].url)
+edit = client.images.edit(model="p-image-edit", prompt="add a red flag",
+                          image="https://example.com/in.png")
+
+# Video (only documented model: p-video, ≤10s)
+vid = client.videos.generate(model="p-video", prompt="ocean waves", duration=5)
+print(vid.url)
+
+# Embeddings — EXPERIMENTAL passthrough (no Qubrid spec; may change)
+e = client.embeddings.create(model="BAAI/bge-large-en-v1.5", input="hello world")
+print(e.data[0].embedding[:4])
+```
+
+More in [`examples/`](examples/).
+
+## Configuration
+
+```python
+QubridClient(
+    api_key="qb_...",              # or QUBRID_API_KEY env
+    base_url="https://platform.qubrid.com/v1",                 # override canonical
+    legacy_base_url="https://platform.qubrid.com/api/v1/qubridai",  # STT only
+    timeout=60.0,                  # seconds; per-call `timeout=` also supported
+    max_retries=2,                 # retry-with-backoff on 429/5xx
+)
+```
+
+## Errors
+
+```python
+from qubrid import QubridError, AuthenticationError, RateLimitError
+
+try:
+    client.chat.completions.create(model="m", messages=[])
+except AuthenticationError:
+    ...  # bad/missing key (401)
+except RateLimitError as e:
+    print(e.status_code, e.code, e.request_id, e.body)
+except QubridError as e:
+    ...  # base class: InvalidRequestError, NotFoundError,
+         # PermissionDeniedError, QuotaExceededError, ServerError, APIConnectionError
+```
+
+## Docs & flags
+
+- STT lives on the legacy prefix — the client handles it; only override `legacy_base_url` for staging/mocks.
+- Live-verified 2026-09-08: runnable examples' `Hunyuan/Hunyuan-OCR-1B` returns 404 `model_not_found`; the working id is `tencent/HunyuanOCR` (the SDK default).
+- Video quickstart JS reads `data[0].video_url` while the spec says `url` — both accepted (`VideoData.resolved_url`).
+- Image-edit has no OpenAPI file (quickstarts only); local files → multipart, remote URLs → JSON `images:[...]`.
+- Anything under `api-reference/endpoint/` (`/plants`) is a Mintlify template placeholder, not a real API.
+
+## Known environment quirks
+
+- **Garbled unicode on Windows terminals:** model output containing emoji, accented characters, or other non-ASCII text may render as `�` in default Windows consoles (e.g. `cmd.exe` / PowerShell with a legacy code page). This is a console-encoding artifact, not an SDK bug — the returned strings are correct. If affected, run with `PYTHONIOENCODING=utf-8` or use a UTF-8-capable terminal (e.g. Windows Terminal).

qubrid-0.1.0/examples/audio.py

@@ -0,0 +1,19 @@
+"""TTS + STT examples."""
+
+from qubrid import QubridClient
+
+client = QubridClient()
+
+speech = client.audio.speech.create(
+    model="qwen3-tts-flash",
+    text="Today is a wonderful day to build something people love!",
+    voice="Cherry",
+    language_type="Auto",
+)
+print(speech.url)
+
+transcript = client.audio.transcriptions.create(
+    model="openai/whisper-large-v3",
+    file="clip.wav",  # local path, file object, or (filename, bytes) tuple
+)
+print(transcript.text)

qubrid-0.1.0/examples/images_video_embeddings.py

@@ -0,0 +1,27 @@
+"""Images, video, embeddings (experimental) examples."""
+
+from qubrid import QubridClient
+
+client = QubridClient()
+
+img = client.images.generate(
+    model="p-image",
+    prompt="a lighthouse at dusk",
+    aspect_ratio="16:9",
+    output_format="webp",
+)
+print(img.data[0].url)
+
+edit = client.images.edit(
+    model="p-image-edit",
+    prompt="add a red flag",
+    image="https://example.com/in.png",  # remote URL → JSON; local path → multipart
+)
+print(edit.data[0].url)
+
+vid = client.videos.generate(model="p-video", prompt="ocean waves at sunset", duration=5)
+print(vid.url)
+
+# EXPERIMENTAL: no Qubrid OpenAPI spec for embeddings; OpenAI-compatible passthrough.
+emb = client.embeddings.create(model="BAAI/bge-large-en-v1.5", input="hello world")
+print(emb.data[0].embedding[:4])

qubrid-0.1.0/examples/quickstart_chat.py

@@ -0,0 +1,27 @@
+"""Qubrid SDK quickstart — chat (non-streaming + streaming)."""
+
+from qubrid import QubridClient
+
+client = QubridClient()  # QUBRID_API_KEY
+
+resp = client.chat.completions.create(
+    model="openai/gpt-oss-120b",
+    messages=[
+        {"role": "user", "content": "Summarize this support ticket into bullet-point next steps."}
+    ],
+    max_tokens=4096,
+    temperature=0.7,
+    top_p=1,
+)
+print(resp.content)
+
+for chunk in client.chat.completions.create(
+    model="openai/gpt-oss-120b",
+    messages=[{"role": "user", "content": "Write a haiku about GPUs."}],
+    stream=True,
+):
+    # Some SSE events carry no text (e.g. role-only first chunk) — skip them
+    # so the output doesn't contain literal "None"s.
+    if chunk.delta:
+        print(chunk.delta, end="", flush=True)
+print()

qubrid-0.1.0/examples/vision_ocr.py

@@ -0,0 +1,18 @@
+"""Vision + OCR examples (docs-example payloads)."""
+
+from qubrid import QubridClient
+
+client = QubridClient()
+
+vision = client.vision.analyze(
+    model="Qwen/Qwen3-VL-Plus",
+    prompt="What is in this image? Describe the main elements.",
+    image_urls="https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg",
+)
+print(vision.content)
+
+ocr = client.ocr.extract(
+    image_urls="https://example.com/receipt.jpg",  # replace with a real image URL
+    # model="tencent/HunyuanOCR",  # default (live-verified)
+)
+print(ocr.content)

qubrid-0.1.0/pyproject.toml

@@ -0,0 +1,64 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "qubrid"
+version = "0.1.0"
+description = "First-party Python SDK for Qubrid AI GPU-accelerated inference (chat, vision, OCR, TTS, STT, images, video, embeddings)"
+readme = "README.md"
+requires-python = ">=3.9"
+license = { text = "MIT" }
+authors = [{ name = "Qubrid AI Community SDK", email = "support@qubrid.com" }]
+keywords = ["qubrid", "ai", "llm", "inference", "sdk", "chat", "vision", "tts", "stt"]
+classifiers = [
+  "Development Status :: 4 - Beta",
+  "Intended Audience :: Developers",
+  "License :: OSI Approved :: MIT License",
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3.9",
+  "Programming Language :: Python :: 3.10",
+  "Programming Language :: Python :: 3.11",
+  "Programming Language :: Python :: 3.12",
+  "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+dependencies = [
+  "httpx>=0.27,<1",
+  "pydantic>=2.0",
+]
+
+[project.urls]
+Homepage = "https://platform.qubrid.com"
+Documentation = "https://github.com/QubridAI-Inc/docs"
+Repository = "https://github.com/QubridAI-Inc/qubrid-examples"
+Changelog = "https://github.com/Abhi-shek26/python-sdk-buildup/blob/main/CHANGELOG.md"
+
+[project.optional-dependencies]
+dev = [
+  "pytest>=8",
+  "pytest-asyncio>=0.23",
+  "respx>=0.22",
+  "mypy>=1.0",
+  "ruff>=0.4",
+]
+
+[tool.hatch.build.targets.wheel]
+packages = ["src/qubrid"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+asyncio_mode = "auto"
+
+[tool.ruff]
+target-version = "py39"
+line-length = 100
+
+[tool.ruff.lint]
+# Optional[...] kept deliberately: `X | None` breaks at runtime on Python 3.9
+# for pydantic-evaluated annotations, so silence pyupgrade rewrites.
+ignore = ["UP006", "UP007", "UP035", "UP045", "PYI034"]
+
+[tool.mypy]
+python_version = "3.9"
+strict = false
+ignore_missing_imports = true