Fix interface + OpenAI URL compatibility + automate all test sets in one go

Author: Sela Tachnai

README.md

@@ -12,6 +12,7 @@ A multi-platform, reproducible tool for benchmarking the **local** energy consum
 
 - 🤖 **Local LLM Support**
   - **Ollama**: Connect to your local Ollama instance
+  - **OpenAI-Compatible**: Connect to vLLM, AnythingLLM, or other OpenAI-compatible APIs
   - More providers coming soon (llama.cpp, etc.)
 
 - 📊 **Comprehensive Metrics**
@@ -111,6 +112,29 @@ Options:
   --help                    Show this message and exit
 
 The runner looks for `testset_<name>.json` inside `src/data/testsets/`. Passing `easy`, `testset_easy`, or the exact filename stem all point to the same file.
+
+### OpenAI-Compatible Providers (vLLM, AnythingLLM, etc.)
+
+To use an OpenAI-compatible provider instead of Ollama:
+
+```bash
+python src/main.py run-test \
+  --model "meta-llama/Meta-Llama-3-8B-Instruct" \
+  --provider openai \
+  --base-url "http://localhost:8000/v1" \
+  --api-key "sk-optional"
+```
+
+### Automation Script
+
+To run all 4 standard test sets (easy, medium, hard, math) in sequence:
+
+```bash
+python run_all_tests.py --model llama3:latest
+```
+
+This will generate output files with the format `results/output_{model}_{dataset}_{date}.json`.
+
 ```
 
 ## Configuration