Platform · served now
One API for chat, embeddings, images and audio — OpenAI-compatible, served from EU regions on hardware AxForge owns, with zero prompt retention.
Catalogue
| Model | Type | Context | Pricing |
|---|---|---|---|
| Qwen3.8 27B | LLM — chat, tools, streaming, usage | 65,536 | €0.29 / 1M in · €1.77 / 1M out (launch pricing) |
| Qwen3 Embedding | Embeddings — 1024-dim vectors | 32,768 | €0.015 / 1M |
| ERNIE Image Turbo | Image generation — dense in-image text | — | €0.02 / image |
| FLUX.2 Klein 4B | Image editing — instruction-based, ~8 s | — | €0.02 / image |
| MiniMax Music 3 | Music generation — full songs to 5 min | — | €0.09 / track |
| Whisper | Speech-to-text — transcription endpoint | — | €0.005 / minute |
| Piper | Text-to-speech — fast CPU TTS | — | €2.95 / 1M characters |
Our pricing rule: we target 80% of the lowest current price from a proper inference supplier for the same model, tracked continuously against the market. Found better pricing at a proper inference supplier? Tell us and we’ll look into lowering ours.
Further models — Qwen3.6 35B A3B, Qwen3 30B A3B, Gemma-4 26B, Mistral Small 3.2, SDXL — are validated on AxForge hardware and deployable on request on a dedicated system. See the model catalogue.
Quickstart
$ export OPENAI_BASE_URL=https://api.axforge.ai/v1 $ curl -sS "$OPENAI_BASE_URL/chat/completions" \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -H "x-axforge-region: eu-se-1" \ -H "Content-Type: application/json" \ -d '{"model":"qwen3.8-27b-nvfp4","stream":true, "messages":[{"role":"user","content":"Summarize DORA in one paragraph."}]}'
# pip install openai from openai import OpenAI client = OpenAI( base_url="https://api.axforge.ai/v1", api_key="YOUR_AXFORGE_KEY", default_headers={"x-axforge-region": "eu-se-1"}, ) r = client.chat.completions.create( model="qwen3.8-27b-nvfp4", messages=[{"role": "user", "content": "Hello from Stockholm"}], ) print(r.choices[0].message.content, r.usage)
The region is pinned on the key; the x-axforge-region header selects it and the response echoes it back, so every call is verifiable.
Data & privacy
Prompts and completions are processed in memory in Sweden — not written to disk, not logged, not retained, and never used to train anything. We keep only request metadata (token counts, timestamps, status) for billing and operations. TLS 1.3 terminates in the EU. GDPR compliant — the full policy is at axforge.ai/privacy.
FAQ
Yes. Chat completions, streaming, tool calls and usage reporting follow the
OpenAI API shape, alongside embeddings, image generation and audio
(transcription and speech). The official OpenAI SDKs work by changing
base_url to https://api.axforge.ai/v1.
Served now: Qwen3.8 27B (chat, tools, streaming), Qwen3 Embedding (1024-dim vectors), ERNIE Image Turbo (image generation), FLUX.2 Klein 4B (image editing), MiniMax Music 3 (music generation), Whisper and Piper. Further models are validated on AxForge hardware and deployable on request on a dedicated system.
In EU regions on hardware AxForge owns: eu-se-1 (Stockholm) and
eu-es-1 (Málaga) are live, with more EU regions in deployment.
Inference runs on NVIDIA DGX Spark (GB10) systems in Sweden, and TLS 1.3
terminates in the EU.
Qwen3.8 27B is €0.29 per million input tokens and €1.77 per million output tokens (launch pricing). Image generation and editing are €0.02 per image, embeddings €0.015 per million tokens, speech-to-text €0.005 per minute, text-to-speech €2.95 per million characters and music €0.09 per track — the full table is above, and the price list has it all.
No. Prompts and completions are processed in memory in Sweden — not written to disk, not logged, not retained, never used to train. Only request metadata (token counts, timestamps, status) is kept for billing and operations. See the privacy policy.
Yes. The region is pinned on the API key and set with the
x-axforge-region header (e.g. eu-se-1); the response
echoes the region back.
Related