California Institute of Technology (Caltech): GPT-OSS 120B, RAG Pipeline
NuClide Research · 2026-05-02
Summary
A Caltech node (yertle.caltech.edu, 131.215.141.46) exposes Ollama with 6 models including gpt-oss:120b (OpenAI’s 120B open model, 65.4GB) and a RAG pipeline stack (two embedding models). The hostname yertle references the Dr. Seuss turtle, a common playful server naming convention at research universities.
Infrastructure
| Field | Value |
|---|---|
| IP | 131.215.141.46 |
| Hostname | yertle.caltech.edu |
| Org | California Institute of Technology |
| Country | United States |
| Ollama version | 0.12.10 |
| Open port | 11434 (public) |
Models
| Model | Size | Notes |
|---|---|---|
| gpt-oss:120b | ~65 GB | OpenAI open-weight, 120B params |
| rjmalagon/gte-qwen2-1.5b-instruct-embed-f16:latest | ~3 GB | Embedding model (RAG) |
| mxbai-embed-large:latest | ~670 MB | Embedding model (RAG) |
| syntax:latest | , | Custom model |
| java:latest | , | Custom model |
| smollm2:135m | 270 MB | SmolLM |
System prompt on smollm2: You are a helpful AI assistant named SmolLM, trained by Hugging Face
The custom syntax:latest and java:latest models suggest research tooling, possibly code syntax assistance or Java-specific fine-tunes.
Findings
F1: 120B Model Accessible for Free Inference (HIGH)
gpt-oss:120b, OpenAI’s 120B open-weight model, accessible without authentication. Frontier-class inference at Caltech’s compute cost.
F2: Active RAG Pipeline (MEDIUM)
Two embedding models (gte-qwen2-1.5b and mxbai-embed-large) co-deployed with a custom syntax:latest model suggests an active RAG pipeline under development, likely for code understanding research.
F3: Model Injection (CRITICAL)
CVE-2025-63389 applies. Custom research models (syntax, java) injectable via /api/create.
Remediation
OLLAMA_HOST=127.0.0.1:11434
systemctl restart ollamaDisclosure
- Discovered: 2026-05-02
- Status: Pending outreach to Caltech IMSS security