GPT-OSS — 120B
Large open-source model, strong in legal and medical analysis.
via ZelixAI Privacy Cluster →
What is this model?
GPT-OSS 120B is a 120-billion-parameter open-source language model with a GPT architecture. Trained on a broad corpus that leans heavily on technical texts, legal documents and scientific literature, it offers a deeper knowledge base for analysis-heavy tasks than smaller models. By making it available within the ZelixAI Privacy Cluster, we deliver flagship-tier reasoning without your data having to leave the European Union.
Strengths
Strengths: the model excels at legal analysis (contract review, case-law comparison), medical text analysis (no diagnosis, but literature and guidelines) and long-document summarisation (policy notes, annual reports). The larger parameter count delivers more accurate answers on complex questions — especially when context is well-prepared via our RAG layer.
Best suited for
- Legal analysis and contract review
- Medical text analysis (not for diagnosis)
- Long documents and summaries
How ZelixAI uses this model
We position GPT-OSS 120B within the Privacy Cluster as the heavier option for analysis work: contract-review bots, policy assistants and long-document summarisation. For real-time customer service Mistral Small is the recommended pick (faster). You switch between models per bot per conversation in the ZelixAI portal — no restart, no integration changes.
Real-world examples within ZelixAI
Concrete praktijkvoorbeelden voor dit model worden binnenkort hier gepubliceerd. Stel intussen vragen via onze contactpagina — we delen graag relevante use-cases uit onze klantbasis.
Limitations and caveats
Limitations: GPT-OSS 120B is significantly slower than Mistral Small (15–25 tokens/sec vs. 60–100). Not suited for real-time chat with immediate-response expectations. Knowledge cut-off is early 2024 — always use our RAG layer for current company data. For matters with legal validity: treat the answer as a starting point, not advice, and always have a human expert verify.
Technical specifications
| Provider | ZelixAI Privacy Cluster |
| Context window | 131K tokens |
| Throughput | 15–40 tokens/s (Average) |
| Cost tier | Very affordable |
| Tool / function-calling | yes |
| Data residency | EU (Netherlands · Germany · France) |