fusion-runtime 0.1.1__tar.gz

fusion_runtime-0.1.1/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 [yyyy] [name of copyright owner]
+
+   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.

fusion_runtime-0.1.1/NOTICE

@@ -0,0 +1,13 @@
+fusion-runtime
+Copyright 2026 Samarth Urs
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use
+this software except in compliance with the License. A copy is in the LICENSE
+file beside this one, and at http://www.apache.org/licenses/LICENSE-2.0
+
+This product bundles no third-party source code. It downloads models at run time
+and installs libraries from PyPI; those carry their own licences, which the
+`frun models list` output and the dependency metadata state. The models used by
+default are permissively licensed (Whisper MIT, Silero VAD MIT, Qwen2.5
+Apache-2.0, Kokoro Apache-2.0), but a model you point it at yourself may not be —
+check before you ship one.

fusion_runtime-0.1.1/PKG-INFO

@@ -0,0 +1,357 @@
+Metadata-Version: 2.4
+Name: fusion-runtime
+Version: 0.1.1
+Summary: Self-hosted voice agent runtime: speech-to-text, LLM and text-to-speech on your own GPU, with tools
+Author-email: Samarth Urs <hello@fusion-runtime.dev>
+License-Expression: Apache-2.0
+Project-URL: Homepage, https://fusion-runtime.dev
+Project-URL: Documentation, https://fusion-runtime.dev/docs
+Project-URL: Repository, https://github.com/SamarthUrs18/fusion-runtime
+Project-URL: Issues, https://github.com/SamarthUrs18/fusion-runtime/issues
+Requires-Python: <3.14,>=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+License-File: NOTICE
+Requires-Dist: fastapi>=0.110.0
+Requires-Dist: uvicorn[standard]>=0.29.0
+Requires-Dist: pydantic>=2.7.0
+Requires-Dist: pydantic-settings>=2.3.0
+Requires-Dist: numpy>=1.26.0
+Requires-Dist: scipy>=1.13.0
+Requires-Dist: soundfile>=0.12.1
+Requires-Dist: websockets>=12.0
+Requires-Dist: python-multipart>=0.0.9
+Requires-Dist: httpx>=0.27.0
+Requires-Dist: python-dotenv>=1.0
+Requires-Dist: faster-whisper>=1.1.0
+Requires-Dist: llama-cpp-python>=0.2.80
+Requires-Dist: onnxruntime>=1.18.0
+Requires-Dist: kokoro-onnx<0.7,>=0.6.1
+Requires-Dist: torch==2.9.1
+Requires-Dist: torchaudio==2.9.1
+Requires-Dist: huggingface-hub>=0.23.0
+Requires-Dist: typer>=0.12
+Requires-Dist: prometheus-client>=0.20
+Requires-Dist: psutil>=5.9
+Provides-Extra: cuda
+Requires-Dist: onnxruntime-gpu<1.30,>=1.18.0; extra == "cuda"
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0.0; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.23.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.1.0; extra == "dev"
+Requires-Dist: ruff>=0.4.0; extra == "dev"
+Requires-Dist: mypy>=1.10.0; extra == "dev"
+Requires-Dist: pre-commit>=3.6.0; extra == "dev"
+Provides-Extra: talk
+Requires-Dist: sounddevice>=0.4.6; extra == "talk"
+Provides-Extra: examples
+Requires-Dist: fusion-runtime[talk]; extra == "examples"
+Provides-Extra: all
+Requires-Dist: fusion-runtime[cuda]; extra == "all"
+Requires-Dist: fusion-runtime[dev]; extra == "all"
+Requires-Dist: fusion-runtime[talk]; extra == "all"
+Dynamic: license-file
+
+<p align="center">
+  <picture>
+    <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/SamarthUrs18/fusion-runtime/main/docs/brand/logo-dark.png">
+    <img alt="fusion-runtime" src="https://raw.githubusercontent.com/SamarthUrs18/fusion-runtime/main/docs/brand/logo-light.png" width="340">
+  </picture>
+</p>
+
+<p align="center">
+  <a href="https://github.com/SamarthUrs18/fusion-runtime"><img alt="GitHub stars" src="https://img.shields.io/github/stars/SamarthUrs18/fusion-runtime?style=social&cacheSeconds=3600"></a>
+  <a href="https://pypi.org/project/fusion-runtime/"><img alt="PyPI" src="https://img.shields.io/pypi/v/fusion-runtime?cacheSeconds=3600"></a>
+  <a href="https://fusion-runtime.dev/docs"><img alt="Docs" src="https://img.shields.io/badge/docs-fusion--runtime.dev-d9612f"></a>
+  <img alt="Python 3.11 to 3.13" src="https://img.shields.io/badge/python-3.11%20%7C%203.12%20%7C%203.13-blue">
+</p>
+
+**A self-hosted voice agent runtime.** Speech-to-text, the LLM and text-to-speech run together on
+one machine and stream into each other, so a reply starts playing while it's still being generated.
+
+**On an RTX 3090 with a 7B model: about 490 ms of processing once a turn ends**, or 991 ms
+stopwatched from your last syllable — the difference is a silence wait you can configure. 127
+tokens/sec, interruptions honoured mid-sentence.
+
+## Quickstart
+
+Requires Python 3.11–3.13.
+
+```bash
+pip install fusion-runtime
+```
+
+An agent is one file. This is the whole thing:
+
+```python
+# agent.py
+from fusion_runtime import Agent, LLM, STT, TTS, Turns
+
+agent = Agent(
+    name="shopkart-orders",
+    prompt="You are the order line for ShopKart. Keep answers to one short sentence.",
+    stt=STT("whisper-tiny.en"),          # or "whisper-small" for better accuracy
+    llm=LLM("qwen2.5-0.5b-q4", max_tokens=256),
+    tts=TTS("kokoro-v1.0", voice="af_heart"),
+    turns=Turns(wait_ms=500, interrupt_after_ms=300),
+)
+```
+
+```bash
+frun models pull agent.py     # exactly the models it names, nothing else
+frun up agent.py              # add --reload to restart on every edit
+```
+
+Then talk to it from a second terminal:
+
+```bash
+pip install "fusion-runtime[talk]"
+frun talk
+```
+
+That is the whole loop — one file, two commands, a conversation. Talk over the agent to
+interrupt it.
+
+`frun up` with no file runs a default agent if you just want to hear it work, and `frun doctor`
+checks libraries, GPU, models and audio and says how to fix what it finds.
+
+### In a browser instead
+
+The runtime serves a browser client at **http://localhost:8000** — the same one you would embed
+in your own page.
+
+With no keys configured, open it and click Talk. With keys configured (`FUSION_ACCEPTED_KEYS`),
+a page can't hold a secret, so it needs a short-lived session token:
+
+```bash
+frun token        # prints a URL with a token in it — open that
+```
+
+Tokens are single-use and expire in about a minute. The page is handed its next one over the
+socket it already has, so a conversation keeps going without asking again. If you open the bare
+URL on a server with keys, the connection closes and the page says the token wasn't accepted.
+
+### Naming models
+
+A model is a catalog id (`frun models list`), a file path, `hf:owner/repo` for anything on
+Hugging Face, or a URL for an OpenAI-compatible endpoint. A model on a vLLM, SGLang or
+llama-server you started is named with that server in front — `vllm:hf:Qwen/Qwen2.5-7B-Instruct-AWQ`
+— and found at the server's usual address (`url=` for another).
+
+Settings are checked against the runtime that runs the model, so a misspelt one fails at startup
+with the name it probably meant instead of being ignored. For llama.cpp that includes
+`flash_attn`, `kv_cache_type="q8_0"` (half the context memory), `use_mlock`, `main_gpu` and
+`llama_kwargs` for anything else; for an endpoint, `extra_body` for server-specific sampling.
+
+Secrets never go in the agent file — it names the *variable* holding a key
+(`api_key_env="GROQ_API_KEY"`), so `agent.py` is safe to commit.
+
+### Tools
+
+A tool is a function. The model reads its name, docstring and type hints, calls it when it needs
+to, and answers with what it returned:
+
+```python
+from fusion_runtime import Agent, LLM, tool
+
+@tool
+async def order_status(order_id: str) -> dict:
+    """Look up where an order is and when it will arrive.
+
+    Args:
+        order_id: The order number, as the caller reads it out.
+    """
+    return await orders.lookup(order_id)
+
+agent = Agent(prompt="...", llm=LLM("vllm:hf:Qwen/Qwen2.5-7B-Instruct-AWQ"), tools=[order_status])
+```
+
+What the model says before calling ("Let me check.") is spoken while the tool runs. Each call has
+a timeout (`@tool(timeout_s=...)`, 10 s by default); ordinary functions run on a worker thread;
+a tool that fails tells the model what went wrong rather than ending the call; and talking over
+the wait cancels it. After `max_tool_rounds` calls in one turn (4) the model has to answer.
+
+Tools need an LLM server that can call them — vLLM (`--enable-auto-tool-choice
+--tool-call-parser ...`), SGLang, llama-server (`--jinja`) or a hosted API. The in-process
+llama.cpp runtime can't, and `frun up` says so at startup. A runnable version is
+[`examples/tools_agent.py`](examples/tools_agent.py).
+
+## On your own site
+
+```html
+<script src="https://your-server/fusion-runtime.js"></script>
+<button id="talk"></button>
+<script>FusionRuntime.attach({ button: "#talk" });</script>
+```
+
+The runtime serves the browser client it uses itself, so the page you demo with is the one your
+site embeds. With no `url` it connects back to wherever the script came from.
+
+Browsers only allow a microphone on `https://`, so a deployment needs TLS and `wss://`. A page
+never holds an API key: your backend mints it a short-lived token.
+
+## Authentication
+
+```bash
+frun key new
+FUSION_ACCEPTED_KEYS=web:frun_kR7m...
+```
+
+Without keys the server answers on `localhost` only, and `frun up --host 0.0.0.0` refuses to
+start. `frun talk`, a backend or curl send the key in an `Authorization` header; a browser page
+gets a short-lived, single-use token from `POST /v1/sessions` instead, because a page can hold
+neither a secret nor a header.
+
+Concurrency caps, message and audio limits, idle timeouts, origin allowlists and proxy trust all
+have working defaults — see the docs.
+
+## Performance
+
+Measured, not estimated. The production profile as it ships — RTX 3090, Qwen 7B q4 + Whisper
+small + Kokoro on the one card — through the browser client, 21 September 2026:
+
+| | Median | Range |
+|---|---|---|
+| **Processing** — turn ends, audio comes back | **~490 ms** | 288–657 |
+| **Stopwatch from your last syllable** | **991 ms** | 858–1061 |
+| ↳ of which: silence wait before the turn is judged over | ~500 ms | `turns.wait_ms` |
+| Speech-to-text | 119 ms | 58–329 |
+| LLM first token | 27 ms | 20–70 |
+| First token → first audio (a sentence gets written, then spoken) | 430 ms | 320–509 |
+| Text-to-speech real-time factor | 0.09 | speech is synthesized ~11× faster than real time |
+| LLM tokens/sec | 127 | 106–130 |
+
+**Two numbers, because there are two honest answers.** A stopwatch started at your last syllable
+reads 991 ms. About 500 ms of that is the runtime waiting through silence to decide you've
+finished — which elapses while you're still finishing, so people don't experience it as waiting.
+What a caller feels is closer to the 490 ms of processing. Quote whichever you like, but say
+which one: a voice stack claiming a number under 500 ms is almost always measuring from "we
+decided the caller stopped", not "the caller stopped".
+
+**The stages don't sum, and that's not sleight of hand.** Transcription of what you already said
+runs during the silence wait. And "first token → first audio" is mostly the language model
+writing a sentence — text-to-speech can't start on half a clause — so it is not a measure of how
+fast Kokoro is. Kokoro's own speed is the real-time factor: 0.09, or about 126 ms of compute for
+1.4 seconds of speech.
+
+Every figure is the runtime's own per-turn telemetry (`frun talk --verbose`, or the browser
+console), so you can reproduce them rather than trusting ours. Barge-in fired on every attempt.
+
+## Several callers at once
+
+Measured on the same 3090, real WebSocket sessions, three turns each
+([`scripts/concurrency_check.py`](scripts/concurrency_check.py)). Response is the server's own
+figure: the turn ends, audio comes back.
+
+| Callers | In-process llama.cpp | vLLM | SGLang |
+|---|---|---|---|
+| 1 | ~460 ms | 398 ms | 410 ms |
+| 4 | ~740 ms | 698 ms | 848 ms |
+| 8 | ~4600 ms | 1086 ms | 1054 ms |
+| 12 | ~7500 ms | 1083 ms | 1264 ms |
+| 16 | — | 2064 ms | 1598 ms |
+
+**With the model in-process, four callers is the ceiling.** At twelve, the language model's first
+token takes 4790 ms of a 5312 ms response, while speech-to-text stays at 76 ms and text-to-speech
+at 469 ms: one llama.cpp context decodes one reply at a time.
+
+**With vLLM or SGLang, twelve callers answer in about a second.** Both batch every caller's reply
+into each step on the GPU, and the language model's first token stayed between 36 and 72 ms from
+one caller to sixteen. The limit moves to speech, which still runs one caller at a time: Kokoro's
+first audio grows from ~350 ms alone to 1.5–2 s at sixteen callers, and Whisper grows with how
+long people talk (a 3-second question: ~200 ms alone, 1.6–2.2 s at sixteen). Batching speech is
+the next step, not the language model.
+
+**Tools hold up under load.** Every caller asked about the same order on every turn: vLLM looked
+it up in 122 of 123 turns, SGLang in 109. On SGLang every session looked it up the first time;
+the turns without a lookup were the question asked a second, third or fourth time, answered from
+the lookup already in the conversation. That is correct, but if your data can change during a
+call, tell the agent to look things up again.
+
+Measured 26 September 2026: Qwen2.5-7B-Instruct-AWQ, `--gpu-memory-utilization 0.6`,
+`--max-model-len 4096`, Whisper small and Kokoro on the same card, both servers with default
+settings otherwise. The in-process column is Qwen 7B q4 GGUF on llama.cpp, 21 September.
+
+To run the model on a server and leave speech where it is:
+
+```python
+llm = LLM("vllm:hf:Qwen/Qwen2.5-7B-Instruct-AWQ", url="http://localhost:8002/v1")
+llm = LLM("llama_server:qwen2.5-7b-instruct")              # llama-server -np N, on :8080
+llm = LLM("http://gpu-box:8000/v1", model_name="...")        # any OpenAI-compatible server
+```
+
+**On one 24 GB card** (RTX 3090, L4) the server shares the GPU with Whisper and Kokoro, and vLLM
+reserves 90% of the card by default. Cap it, and start it first:
+
+```bash
+vllm serve Qwen/Qwen2.5-7B-Instruct-AWQ --port 8002 \
+  --gpu-memory-utilization 0.6 --max-model-len 4096 --max-num-seqs 16 \
+  --enable-auto-tool-choice --tool-call-parser hermes
+frun up agent.py        # an Agent(profile="production", ...), so Whisper runs on the GPU too
+```
+
+SGLang the same way (its usual port is 30000, and `sglang:` finds it there):
+
+```bash
+python -m sglang.launch_server --model-path Qwen/Qwen2.5-7B-Instruct-AWQ --port 30000 \
+  --mem-fraction-static 0.6 --context-length 4096 --tool-call-parser qwen25
+```
+
+`0.6` is about 14 GB: the weights plus every caller's context. Voice turns are short, so a 4096
+context fits more callers than the model's maximum would. The tool flags are only needed for
+tools, and the parser depends on the model family. Port 8002, because `frun up` is on 8000 and some hosts (Runpod's pod images) already use 8001.
+
+A 4-bit model (AWQ or GPTQ) leaves room for speech; a 16-bit 7B model needs ~15 GB for its
+weights alone and doesn't. The L4 has about a third of the 3090's memory bandwidth, so expect
+slower tokens there. `nvidia-smi` shows what is actually used.
+
+## The `frun` CLI
+
+| | |
+|---|---|
+| `frun up [agent.py]` | Starts the server. `--host`, `--port`, `--reload`, `--config` |
+| `frun talk` | Talks to it from a terminal, with a latency summary per turn |
+| `frun models list` / `pull` | What's available, and downloading it |
+| `frun key new` / `keys list` / `token` | Keys and browser tokens |
+| `frun doctor` | Checks the machine and says how to fix what's wrong |
+| `frun version` | The installed version. `--version` and `-V` work too |
+
+`fusion-runtime` works as an alias for `frun`.
+
+## Using it as a library
+
+`frun up agent.py` covers running an agent. The pipeline can also run inside your own process —
+for a queue worker, a test, or a batch job over recorded calls — with no server involved. See
+[`examples/sdk_example.py`](examples/sdk_example.py), which is runnable, and
+[the docs](https://fusion-runtime.dev/docs#python).
+
+## Documentation and contact
+
+Everything else — configuration, turn detection, languages, the server API, telemetry, limits,
+GPU setup and deployment — is at **[fusion-runtime.dev/docs](https://fusion-runtime.dev/docs)**.
+
+| | |
+|---|---|
+| Site and docs | **[fusion-runtime.dev](https://fusion-runtime.dev)** |
+| Questions, or anything else | **hello@fusion-runtime.dev** |
+| Security problems | **security@fusion-runtime.dev** — not a public issue, please ([why](CONTRIBUTING.md#security)) |
+
+## Development
+
+```bash
+uv sync --extra dev --extra talk     # or: pip install -e ".[dev,talk]"
+pytest
+```
+
+CI runs the suite on Python 3.11, 3.12 and 3.13. [CONTRIBUTING.md](CONTRIBUTING.md) has the
+layout, the design rules a review will hold you to, and how to add a runtime.
+
+## License
+
+[Apache-2.0](LICENSE). Embed it in a commercial product, rebrand it, ship it closed — keep the
+copyright notice and the `NOTICE` file in what you distribute, and don't use the project's name
+to imply it endorses you.
+
+The models it downloads by default are permissive too (Whisper MIT, Silero VAD MIT, Qwen2.5
+Apache-2.0, Kokoro Apache-2.0), so the whole default path is clear for commercial use. A model
+you point it at yourself carries its own licence — check that one before you ship it.