Managed inference for open models — EEA-resident, zero data retention by default.
Sference (operated by Neural Compute Ltd, registered in England and Wales) runs a managed inference platform for open-weight models, built as its own scheduler and runtime rather than a wrapper around a hyperscaler. Inference runs on infrastructure within the EEA, and every model is pinned to a specific version so a silent upstream swap can't change your outputs. On Opper it serves GLM-5.2, Kimi K3, DeepSeek V4 Flash, Qwen 3.6 35B-A3B, Qwen3-VL 30B, and BottleCap AI's reasoning-efficient ThinkingCap. A strong pick when you want European residency for frontier open weights.
Every model we route through Sference. Compare residency, zero data retention, training posture and price at a glance, with full data-handling detail per route below.
| Model | Region | Zero data retention | Training | Context | Input | Output |
|---|---|---|---|---|---|---|
| EU | No | 1M | $1.20 | $4.20 | ||
| EU | No | 1M | $3.00 | $15.00 | ||
| EU | No | 1M | $0.20 | $0.60 | ||
| EU | No | 1M | $0.50 | $1.50 | ||
| EU | No | 1M | $0.28 | $0.56 | ||
| EU | No | 1M | $0.28 | $0.56 | ||
| EU | No | 1M | $0.28 | $0.56 | ||
| EU | No | 1M | $1.20 | $4.20 |
Sference hosts on 1 route. Each route has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.
Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. EEA; SCCs; DPA available.
