Qwen3.8-27B API
Hosted inference for Qwen3.8-27B, served from EU hardware in NVFP4 on Blackwell. OpenAI-compatible endpoint, streaming, prompt caching, 262,144-token context. Every performance number below is measured on live production traffic — see the status page.
| Input | Output | Cached input |
|---|---|---|
| $0.25 | $2.09 | $0.04 |
Prompt prefixes served from cache bill at the cached-input rate — roughly one tenth of the input price. No subscription, no minimums. Volume pricing for sustained load: artem@llmtech.eu.
from openai import OpenAI
client = OpenAI(
base_url="https://api.llmtech.eu/v1",
api_key="YOUR_KEY", # get one: artem@llmtech.eu
)
r = client.chat.completions.create(
model="unsloth/Qwen3.8-27B-NVFP4",
messages=[{"role": "user", "content": "Hello"}],
stream=True,
)
for chunk in r:
print(chunk.choices[0].delta.content or "", end="")
API keys are issued by hand while we are in early production — email artem@llmtech.eu and you will have one the same day. Self-service is planned.
Zero data retention
Prompts and completions are processed in volatile memory only — never written to disk, logs, analytics or backups, and never used for training. We retain per-request technical metadata only: timestamps, token counts, status codes and billing prices.
GDPR compliance and DPA
LLM Tech operates from the EU (Poland) and processes all data on EU hardware. A signable Data Processing Agreement under Art. 28 GDPR is available on request, including our technical and organisational measures and the full sub-processor list (infrastructure only — no sub-processor has access to request content).
Where does inference run?
All compute is located in the European Union: the GPU node is in Helsinki, Finland, and the edge is in Nuremberg, Germany. No data leaves the EU/EEA and there is no US cloud at any hop. TLS 1.3 only, end to end.
How fast is it, really?
Time to first token is dominated by prompt length, so a single number would be misleading. On short prompts we measure around 250 ms. On the 60,000-token prompts that make up most of our live traffic, the 24-hour median sits near 1.7 s, because that time goes into prefill. Generation runs at 60–85 tok/s depending on how many requests share the card. We would rather publish that spread than quote our best case: the status page shows the current numbers, whatever they are today.
How do I verify your performance claims?
The status page publishes uptime, TTFT and generation speed measured on live production traffic, refreshed every 5 minutes — including an hourly uptime strip for the last 24 hours. Nothing on this page is a number we do not also publish there.
Does the model accept images?
Yes. The NVFP4 quantisation keeps the vision tower, so image input works. Image tokens are counted into prompt_tokens and itemised at usage.prompt_tokens_details.multimodal_tokens, billed at the input rate — a 64×64 PNG costs 64 image tokens. Most third-party listings for this model describe it as text-only; that is wrong. Images must go in a user message, not a system message.
How does prompt caching work, and what does it cost?
Repeated prompt prefixes are cached automatically and billed at $0.04 per million instead of $0.25. The cache uses 1,584-token blocks and materialises lazily, so a prefix has to be seen three times before it pays off. Keep the system prompt and tool definitions byte-identical and in a stable order, and put anything volatile after them. On our own production traffic this brings the effective input price to about $0.07 per million.