# Available Models

!!! warning "Models change quickly"

    Our model lineup changes every few months: a model usually stays available for **3–4 months, at most about 6**.
    New models and retirements are announced in the [Release Notes](https://datalab.tuwien.ac.at/aiml/aqueduct/release-notes/) and the
    [Matrix room](https://matrix.to/#/#llm-service:tuwien.ac.at). If your work depends on one specific model,
    for example your thesis, read [How long does a model stay available?](https://datalab.tuwien.ac.at/aiml/faq/#how-long-does-a-model-stay-available) first.

All models are available through the OpenAI-compatible API at `https://aqueduct.ai.datalab.tuwien.ac.at/v1`
and in [Open WebUI](https://datalab.tuwien.ac.at/aiml/openwebui/).

!!! warning "Upcoming deprecations — 24 November 2026"

    `kokoro`, `piper-thorsten` and `z-image-turbo` are scheduled for deprecation on 24 November 2026.

## Language Models

| Model ID | Context | Vision | Reasoning efforts | Limits | Status |
| --- | --- | :---: | --- | :---: | --- |
| `deepseek-v4-vision-exp-284b` | 393,216 | :material-check: | low · high · max | 1× | <span class="badge badge--stable">Stable</span> |
| `glm-5.3-744b-experimental` | 262,144 | | low · high · max | 1× | <span class="badge badge--experimental">Experimental</span> |
| `qwen-3.8-27b` | 262,144 | :material-check: | low · medium · xhigh | 2× | <span class="badge badge--stable">Stable</span> |

**Context** is the maximum usable context window in tokens, including the output. **Limits** is the multiplier on
your personal [rate limits](https://datalab.tuwien.ac.at/aiml/faq/#how-are-rate-limits-calculated-and-when-do-i-get-a-429).

### Model details

`deepseek-v4-vision-exp-284b`
:   [DeepSeek-V4-Flash-Vision-Exp](https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp) on Hugging Face.
    Our main model. 2 instances, each running on 4x NVIDIA RTX PRO 6000 Blackwell.

`glm-5.3-744b-experimental`
:   [GLM-5.3](https://huggingface.co/zai-org/GLM-5.3) on Hugging Face.
    **Experimental deployment**: please only use it for coding or day-to-day tasks. It may be removed at short notice.
    1 instance, running on 8x AMD Instinct MI300X.

`qwen-3.8-27b`
:   [Qwen3.8-27B-FP8](https://huggingface.co/Qwen/Qwen3.8-27B-FP8) on Hugging Face.
    Fast, smaller model, also available under the aliases `vision` and `ocr`.
    4 instances, each running on 1x NVIDIA H200.

### Context and output tokens

The context lengths in the table are the **maximum usable** values: longer inputs are trimmed to that size, and output
tokens count toward the context window. Answers can be up to **65536 tokens** long by default; set `max_tokens` to change
this per request.

Some clients (e.g. Pi, Continue) have their own `maxTokens` setting that cuts answers short. Set it to 65536 to use the
full output budget.

### Reasoning

The language models can think (reason) before they answer. You can control this per request:

- **`reasoning_effort`**: how much the model thinks, with the values from the **Reasoning efforts** column
- **`enable_thinking`**: `false` turns thinking off, `true` turns it on, overriding the server default

With the OpenAI-compatible API, both go in `chat_template_kwargs`:

``` python
# client = OpenAI(...)
response = client.chat.completions.create(
    model=model,
    messages=messages,
    extra_body={
        "chat_template_kwargs": {
            "enable_thinking": True,    # overrides server default
            "reasoning_effort": "medium",
        },
    },
)
```

Some clients (e.g. older versions of certain coding tools) don't pass `reasoning_effort` through. Aqueduct then returns
`aqueduct.UnsupportedParamsError`: update the client or use a model alias without reasoning. For all request-level
options, see [reasoning outputs](https://docs.vllm.ai/en/stable/features/reasoning_outputs/#request-level-override) in
the vLLM documentation.

## Text-to-Speech (TTS)

See the [speech guide](https://tu-wien-datalab.github.io/aqueduct/api/speech/) in the public Aqueduct documentation for usage.

| Model ID | Language | Voices | Sample |
| --- | --- | --- | --- |
| [`voxtral-4b-tts`](https://huggingface.co/mistralai/Voxtral-4B-TTS-2603) | 9 languages | 20 preset voices | — |

`voxtral-4b-tts` supports English, French, Spanish, Portuguese, Italian, Dutch, German, Arabic and Hindi.

**English and German voice presets:**

| Language | Voice presets |
| --- | --- |
| English | `neutral_male`, `neutral_female`, `casual_male`, `casual_female`, `cheerful_female` |
| German | `de_male`, `de_female` |


## Speech-to-Text (STT)

See the [speech guide](https://tu-wien-datalab.github.io/aqueduct/api/speech/) in the public Aqueduct documentation for usage.

| Model ID | Languages |
| --- | --- |
| [`whisper-large`](https://huggingface.co/Systran/faster-whisper-large-v3) | 99 languages, including English and German |

## Image Generation

See the [image generation guide](https://tu-wien-datalab.github.io/aqueduct/api/image_generation/) in the public Aqueduct documentation for usage.

| Model ID |
| --- |
| [`qwen-image-2-1-turbo`](https://huggingface.co/Qwen/Qwen-Image-2.1-Turbo) |
| [`z-image-turbo`](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) |

`qwen-image-2-1-turbo` is the 8-step Turbo checkpoint for text-to-image generation. Aqueduct currently does not support
image-edit endpoints.

## Listing models via the API

The API always returns the current lineup:

``` bash
curl https://aqueduct.ai.datalab.tuwien.ac.at/v1/models \
  -H "Authorization: Bearer $AQUEDUCT_API_KEY" | jq
```

How well each model supports specific API features (streaming, tool calling, structured output, …) is tested
nightly, see the [Model Compatibility](https://datalab.tuwien.ac.at/aiml/aqueduct/model-compatibility/) report.