diff --git a/ADD_SONNET_4.5_INSTRUCTIONS.md b/ADD_SONNET_4.5_INSTRUCTIONS.md
new file mode 100644
index 0000000..171be4b
--- /dev/null
+++ b/ADD_SONNET_4.5_INSTRUCTIONS.md
@@ -0,0 +1,136 @@
+# Adding Claude Sonnet 4.5 to LLM Benchmarks
+
+## Overview
+This document explains how to add the new Claude Sonnet 4.5 model to your LLM benchmarking system.
+
+## Current Status
+- **Release Date**: September 29, 2025
+- **Model Type**: Cloud API-based (not for local deployment)
+- **Provider**: Anthropic
+
+## Model Identifier
+⚠️ **IMPORTANT**: You need to verify the exact model ID from Anthropic's documentation.
+
+Based on existing naming patterns, it's likely one of:
+- `claude-4-sonnet-20250929`
+- `claude-4-5-sonnet-20250929`
+- `claude-4-sonnet-20250514`
+
+Check: https://docs.anthropic.com/en/docs/models-overview
+
+## How to Add the Model
+
+### Prerequisites
+- Access to your MongoDB instance
+- `mongosh` installed on your machine
+- Environment variables set:
+  - `MONGODB_URI`: e.g., `mongodb+srv://user:pass@cluster.mongodb.net`
+  - `MONGODB_DB`: e.g., `llm-bench` or `llmbench_staging`
+
+### Method 1: Add to Database (Recommended for Production)
+
+1. **Verify the correct model ID** from Anthropic's API documentation
+
+2. **Seed the model into MongoDB**:
+   ```bash
+   # Replace MODEL_ID with the actual model identifier
+   PROVIDER=anthropic MODEL_ID=claude-4-sonnet-20250929 mongosh "$MONGODB_URI/$MONGODB_DB" scripts/seed_model.js
+   ```
+
+3. **Verify it was added**:
+   ```bash
+   mongosh "$MONGODB_URI/$MONGODB_DB" --eval 'db.models.find({provider: "anthropic", enabled: true}).pretty()'
+   ```
+
+### Method 2: Update Configuration File
+
+Edit `/workspace/cloud/models.json` and add the new model to the `anthropic` array:
+
+```json
+"anthropic": [
+    "claude-2.1",
+    "claude-3-haiku-20240307",
+    "claude-3-5-haiku-20241022",
+    "claude-3-sonnet-20240229",
+    "claude-3-5-sonnet-20240620",
+    "claude-3-7-sonnet-20250219",
+    "claude-3-opus-20240229",
+    "claude-4-sonnet-20250929"  // <-- ADD THIS (verify exact ID)
+]
+```
+
+**Note**: This method updates the reference configuration. The actual benchmarking system reads from MongoDB.
+
+### Method 3: Enqueue a Job (Optional)
+
+If you want to immediately queue a benchmark job:
+```bash
+PROVIDER=anthropic MODEL=claude-4-sonnet-20250929 IGNORE_FRESHNESS=true mongosh "$MONGODB_URI/$MONGODB_DB" scripts/enqueue_job.js
+```
+
+## Testing the Integration
+
+### 1. Test with headless benchmark (local, no Docker):
+```bash
+python api/bench_headless.py --providers anthropic --limit 1 --fresh-minutes 0
+```
+
+### 2. Check for errors:
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval 'db.errors_cloud.find({provider: "anthropic"}).sort({ts: -1}).limit(5).pretty()'
+```
+
+### 3. View results:
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval 'db.metrics_cloud_staging.find({provider: "anthropic", model_name: "claude-4-sonnet-20250929"}).sort({gen_ts: -1}).limit(1).pretty()'
+```
+
+## Database Schema
+
+Models are stored with the following structure:
+```javascript
+{
+  provider: "anthropic",
+  model_id: "claude-4-sonnet-20250929",
+  enabled: true,
+  added_at: ISODate("2025-09-30T...")
+}
+```
+
+## Code Files Involved
+
+1. **`/workspace/api/llm_bench/models_db.py`**: Loads models from MongoDB
+2. **`/workspace/cloud/models.json`**: Reference configuration
+3. **`/workspace/scripts/seed_model.js`**: Script to add models to DB
+4. **`/workspace/api/llm_bench/cloud/providers/anthropic.py`**: Anthropic API integration
+
+## Required Environment Variables
+
+Ensure you have:
+- `ANTHROPIC_API_KEY`: Your Anthropic API key
+- `MONGODB_URI`: MongoDB connection string
+- `MONGODB_DB`: Database name
+- `MONGODB_COLLECTION_MODELS`: Collection name (default: `models`)
+
+## Next Steps
+
+1. ✅ Find the exact model ID from Anthropic docs
+2. ✅ Ensure `ANTHROPIC_API_KEY` is set in your environment
+3. ✅ Run the seed_model.js script with the correct model ID
+4. ✅ Test with a benchmark run
+5. ✅ Monitor for any errors
+
+## Troubleshooting
+
+**If the model doesn't appear in benchmarks:**
+- Check it's marked as `enabled: true` in the database
+- Verify the model_id matches exactly what Anthropic's API expects
+- Check logs for API errors (might indicate wrong model ID)
+
+**If you get authentication errors:**
+- Verify `ANTHROPIC_API_KEY` is set correctly
+- Check the API key has access to the new model
+
+**If the model runs but fails:**
+- The model ID might be incorrect
+- Check Anthropic's API status/changelog for any breaking changes
\ No newline at end of file
diff --git a/FILES_CREATED.md b/FILES_CREATED.md
new file mode 100644
index 0000000..36437d7
--- /dev/null
+++ b/FILES_CREATED.md
@@ -0,0 +1,290 @@
+# Files Created for Adding Claude Sonnet 4.5
+
+This document lists all the new files created to help you add the Claude Sonnet 4.5 model to your LLM Benchmarks database.
+
+---
+
+## 📚 Documentation Files
+
+### 1. **QUICK_START.md**
+**Purpose**: Fast reference - get started in 30 seconds  
+**Use when**: You just want to add the model quickly
+
+```bash
+cat QUICK_START.md
+```
+
+### 2. **RESEARCH_SUMMARY.md**
+**Purpose**: Complete research findings and all available methods  
+**Use when**: You want to understand how everything works
+
+Includes:
+- How the system works
+- All 4 methods to add models
+- Testing procedures
+- Troubleshooting guide
+- Key files in codebase
+
+### 3. **ADD_SONNET_4.5_INSTRUCTIONS.md**
+**Purpose**: Detailed step-by-step instructions  
+**Use when**: You want comprehensive guidance
+
+Includes:
+- Prerequisites checklist
+- All methods with examples
+- Testing procedures
+- Environment variables
+- Next steps
+
+### 4. **FILES_CREATED.md** (this file)
+**Purpose**: Index of all created files  
+**Use when**: You want to see what was created
+
+---
+
+## 🔧 Utility Scripts
+
+### 5. **scripts/add_model_to_db.py** ⭐ RECOMMENDED
+**Purpose**: Python script to add models to MongoDB  
+**Language**: Python  
+**Prerequisites**: `pymongo` installed
+
+**Features**:
+- ✅ Add any model to the database
+- ✅ Check for duplicates before inserting
+- ✅ Update existing models
+- ✅ List all models
+- ✅ Filter by provider
+- ✅ Interactive confirmation
+- ✅ Verification after insertion
+- ✅ Helpful error messages
+
+**Usage**:
+```bash
+# Add Sonnet 4.5
+python scripts/add_model_to_db.py \
+  --provider anthropic \
+  --model-id claude-4-sonnet-20250929
+
+# List all Anthropic models
+python scripts/add_model_to_db.py --list --provider anthropic
+
+# Add a disabled model
+python scripts/add_model_to_db.py \
+  --provider openai \
+  --model-id gpt-5 \
+  --disabled
+
+# Get help
+python scripts/add_model_to_db.py --help
+```
+
+### 6. **scripts/add_sonnet_4.5.sh**
+**Purpose**: Bash script specifically for adding Sonnet 4.5  
+**Language**: Bash  
+**Prerequisites**: `mongosh` installed
+
+**Features**:
+- ✅ Pre-configured for Anthropic Sonnet 4.5
+- ✅ Validation checks
+- ✅ Warns about default model ID
+- ✅ Verifies insertion
+- ✅ Shows next steps
+
+**Usage**:
+```bash
+# Edit the MODEL_ID in the script first, then:
+bash scripts/add_sonnet_4.5.sh
+```
+
+### 7. **scripts/test_anthropic_models.py** ⭐ VERY USEFUL
+**Purpose**: Test Anthropic API access and discover model IDs  
+**Language**: Python  
+**Prerequisites**: `anthropic` package installed
+
+**Features**:
+- ✅ Verify ANTHROPIC_API_KEY is working
+- ✅ Test connectivity to Anthropic API
+- ✅ Test specific model IDs
+- ✅ List known models
+- ✅ Test all known models
+- ✅ Get actual responses from models
+- ✅ Helpful error messages
+
+**Usage**:
+```bash
+# Test a specific model ID to verify it works
+python scripts/test_anthropic_models.py \
+  --test-model claude-4-sonnet-20250929
+
+# Test all known models
+python scripts/test_anthropic_models.py --test-all-known
+
+# Just show info and known models
+python scripts/test_anthropic_models.py
+```
+
+**This is the best way to find the correct model ID!**
+
+---
+
+## 🗂️ File Structure
+
+```
+/workspace/
+├── QUICK_START.md                      # Quick reference
+├── RESEARCH_SUMMARY.md                 # Complete research
+├── ADD_SONNET_4.5_INSTRUCTIONS.md     # Detailed guide
+├── FILES_CREATED.md                    # This file
+└── scripts/
+    ├── add_model_to_db.py             # Python script (recommended)
+    ├── add_sonnet_4.5.sh              # Bash script
+    ├── test_anthropic_models.py       # Test Anthropic API
+    ├── seed_model.js                  # Original MongoDB script
+    └── ...
+```
+
+---
+
+## 🚀 Recommended Workflow
+
+### Step 1: Find the Correct Model ID
+
+```bash
+# First, test what model IDs work
+python scripts/test_anthropic_models.py \
+  --test-model claude-4-sonnet-20250929
+```
+
+If that fails, try variations:
+- `claude-4-5-sonnet-20250929`
+- `claude-sonnet-4-5-20250929`
+- Check Anthropic's docs
+
+### Step 2: Add to Database
+
+```bash
+# Once you know the correct ID, add it
+python scripts/add_model_to_db.py \
+  --provider anthropic \
+  --model-id <CORRECT_MODEL_ID>
+```
+
+### Step 3: Test Benchmark
+
+```bash
+# Run a test benchmark
+python api/bench_headless.py \
+  --providers anthropic \
+  --limit 1
+```
+
+### Step 4: Monitor
+
+```bash
+# Check for errors
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.errors_cloud.find({
+  provider: "anthropic"
+}).sort({ts: -1}).limit(5).pretty()
+'
+```
+
+---
+
+## 📝 Key Takeaways
+
+### ✅ What You Can Do
+
+1. **Add models via code** - No manual database editing needed
+2. **Test before adding** - Verify model IDs work first
+3. **List existing models** - See what's already in the DB
+4. **Update models** - Change enabled/disabled status
+
+### ✅ What You Need
+
+1. **MongoDB access** - Connection string and credentials
+2. **Anthropic API key** - For testing and benchmarking
+3. **Correct model ID** - From Anthropic's documentation
+4. **Python/mongosh** - To run the scripts
+
+### ✅ What You Don't Need
+
+1. ❌ Local model files - Anthropic is API-only
+2. ❌ GPU/special hardware - API handles compute
+3. ❌ Code changes - Everything is config-driven
+4. ❌ Restart services - MongoDB changes are live
+
+---
+
+## 🆘 If Something Goes Wrong
+
+### Script fails to run
+```bash
+# Check Python dependencies
+pip install pymongo anthropic
+
+# Check mongosh is installed
+mongosh --version
+```
+
+### Can't connect to MongoDB
+```bash
+# Verify connection
+mongosh "$MONGODB_URI" --eval 'db.runCommand({ping: 1})'
+```
+
+### Model test fails
+```bash
+# Check API key
+echo $ANTHROPIC_API_KEY
+
+# Try a known working model first
+python scripts/test_anthropic_models.py \
+  --test-model claude-3-5-sonnet-20240620
+```
+
+### Model not appearing in benchmarks
+```bash
+# Check it's enabled
+python scripts/add_model_to_db.py --list --provider anthropic
+
+# Verify in database
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.models.find({provider: "anthropic", enabled: true}).pretty()
+'
+```
+
+---
+
+## 📖 Additional Resources
+
+- **Anthropic Docs**: https://docs.anthropic.com/en/docs/models-overview
+- **MongoDB Shell**: https://www.mongodb.com/docs/mongodb-shell/
+- **PyMongo Docs**: https://pymongo.readthedocs.io/
+
+---
+
+## 💡 Pro Tips
+
+1. **Always test the model ID first** using `test_anthropic_models.py`
+2. **Use the Python script** (`add_model_to_db.py`) - it's more robust
+3. **List models frequently** to see what's in your database
+4. **Start with one model** - don't add multiple until first one works
+5. **Check the dates** - Model IDs use YYYYMMDD format (20250929 = Sep 29, 2025)
+
+---
+
+## Summary
+
+I've created 7 files for you:
+- **4 documentation files** explaining everything
+- **3 utility scripts** to make adding models easy
+
+**Start here**: 
+1. Read `QUICK_START.md` (30 seconds)
+2. Run `test_anthropic_models.py --test-model <MODEL_ID>`
+3. Run `add_model_to_db.py --provider anthropic --model-id <MODEL_ID>`
+4. Done!
+
+Good luck! 🚀
\ No newline at end of file
diff --git a/QUICK_START.md b/QUICK_START.md
new file mode 100644
index 0000000..799112a
--- /dev/null
+++ b/QUICK_START.md
@@ -0,0 +1,80 @@
+# Quick Start: Add Claude Sonnet 4.5
+
+## 🚀 TL;DR
+
+```bash
+# 1. Get the exact model ID from Anthropic docs
+#    https://docs.anthropic.com/en/docs/models-overview
+
+# 2. Set environment variables (if not already set)
+export MONGODB_URI="your-mongodb-uri"
+export MONGODB_DB="llm-bench"
+export ANTHROPIC_API_KEY="your-api-key"
+
+# 3. Add the model (replace MODEL_ID with actual ID)
+python scripts/add_model_to_db.py \
+  --provider anthropic \
+  --model-id claude-4-sonnet-20250929
+
+# 4. Test it
+python api/bench_headless.py --providers anthropic --limit 1
+```
+
+## ✅ That's it!
+
+---
+
+## 📋 Checklist
+
+- [ ] Get exact model ID from Anthropic
+- [ ] Set `MONGODB_URI` environment variable
+- [ ] Set `ANTHROPIC_API_KEY` environment variable
+- [ ] Run the add_model_to_db.py script
+- [ ] Test with a benchmark run
+- [ ] Check for errors
+
+---
+
+## 📚 Full Documentation
+
+For detailed information, see:
+- **RESEARCH_SUMMARY.md** - Complete research and all methods
+- **ADD_SONNET_4.5_INSTRUCTIONS.md** - Step-by-step guide
+- **scripts/add_model_to_db.py** - Python script with `--help`
+
+---
+
+## 🔍 Verify Model ID
+
+The model ID likely follows this pattern:
+```
+claude-4-sonnet-YYYYMMDD
+```
+
+Where YYYYMMDD is the release date (e.g., 20250929 for Sep 29, 2025).
+
+**Check Anthropic's docs to confirm!**
+
+---
+
+## ❓ Common Questions
+
+**Q: Do I need local server access?**  
+A: Only if your MongoDB is running locally. If it's remote, you can run from anywhere.
+
+**Q: Will this work?**  
+A: Yes! The system is designed for this. Just need the correct model ID.
+
+**Q: What if the model ID is wrong?**  
+A: The benchmark will fail with an API error. Update the ID in the database.
+
+**Q: Can I test without adding to database?**  
+A: No, the system reads models from MongoDB only.
+
+---
+
+## 🆘 Need Help?
+
+1. Check `/workspace/RESEARCH_SUMMARY.md` for complete details
+2. Run with `--help`: `python scripts/add_model_to_db.py --help`
+3. List existing models: `python scripts/add_model_to_db.py --list --provider anthropic`
\ No newline at end of file
diff --git a/README_ADDING_MODELS.md b/README_ADDING_MODELS.md
new file mode 100644
index 0000000..4989860
--- /dev/null
+++ b/README_ADDING_MODELS.md
@@ -0,0 +1,358 @@
+# Adding New Models to LLM Benchmarks
+
+**Quick Navigation**: 🚀 [Quick Start](#quick-start) | 📚 [Documentation](#documentation) | 🔧 [Tools](#tools) | ❓ [FAQ](#faq)
+
+---
+
+## 🎯 Goal
+
+Add **Claude Sonnet 4.5** (or any new model) to your LLM Benchmarks database.
+
+## ✅ Answer: Is This Fixable Via Code?
+
+**YES!** The system is designed for this. You can add models through:
+- MongoDB scripts
+- Python scripts (recommended)
+- Direct database commands
+
+**No code changes needed** - it's all configuration-driven.
+
+---
+
+## 🚀 Quick Start
+
+### The Fastest Way (3 commands)
+
+```bash
+# 1. Find the correct model ID
+python scripts/test_anthropic_models.py --test-model claude-4-sonnet-20250929
+
+# 2. Add it to the database
+python scripts/add_model_to_db.py --provider anthropic --model-id claude-4-sonnet-20250929
+
+# 3. Test it
+python api/bench_headless.py --providers anthropic --limit 1
+```
+
+**Done!** 🎉
+
+For more details, see [QUICK_START.md](QUICK_START.md)
+
+---
+
+## 📚 Documentation
+
+I've created comprehensive documentation for you:
+
+| File | Purpose | When to Use |
+|------|---------|-------------|
+| **[QUICK_START.md](QUICK_START.md)** | 30-second guide | You want to add the model NOW |
+| **[RESEARCH_SUMMARY.md](RESEARCH_SUMMARY.md)** | Complete research & all methods | You want to understand everything |
+| **[ADD_SONNET_4.5_INSTRUCTIONS.md](ADD_SONNET_4.5_INSTRUCTIONS.md)** | Step-by-step guide | You want detailed instructions |
+| **[FILES_CREATED.md](FILES_CREATED.md)** | Index of new files | You want to see what was created |
+
+---
+
+## 🔧 Tools
+
+I've created three utility scripts:
+
+### 1. Test Anthropic Models ⭐ START HERE
+
+**Purpose**: Find the correct model ID and verify API access
+
+```bash
+python scripts/test_anthropic_models.py --test-model claude-4-sonnet-20250929
+```
+
+**This answers**: "What's the correct model ID?"
+
+### 2. Add Model to Database ⭐ RECOMMENDED
+
+**Purpose**: Add any model to MongoDB
+
+```bash
+python scripts/add_model_to_db.py --provider anthropic --model-id <MODEL_ID>
+```
+
+**Features**: Duplicate checking, verification, list models, interactive
+
+### 3. Add Sonnet 4.5 (Bash)
+
+**Purpose**: Bash alternative for adding Sonnet 4.5
+
+```bash
+bash scripts/add_sonnet_4.5.sh
+```
+
+**Note**: Edit the script first to set the correct model ID
+
+---
+
+## 🔍 Research Findings
+
+### How Models Are Stored
+
+- **Database**: MongoDB
+- **Collection**: `models` (configurable)
+- **Structure**:
+  ```javascript
+  {
+    provider: "anthropic",
+    model_id: "claude-4-sonnet-20250929",
+    enabled: true,
+    added_at: ISODate(...)
+  }
+  ```
+
+### How Models Are Loaded
+
+The Python code (`api/llm_bench/models_db.py`) queries MongoDB:
+
+```python
+db.models.find({"enabled": True})
+```
+
+This means:
+- ✅ Add models = add to MongoDB
+- ✅ Enable/disable = update `enabled` field
+- ✅ Changes are immediate (no restart needed)
+
+### Current Anthropic Models in Your System
+
+From your database/config:
+- `claude-3-7-sonnet-20250219` ← **Newest currently**
+- `claude-3-5-sonnet-20240620`
+- `claude-3-5-haiku-20241022`
+- `claude-3-opus-20240229`
+- Others...
+
+---
+
+## 🎯 About Claude Sonnet 4.5
+
+### Release Info
+- **Date**: September 29, 2025
+- **Type**: Cloud API only
+- **Capabilities**: Improved coding, finance, cybersecurity, long-duration tasks
+
+### Model Naming Pattern
+
+Anthropic uses: `claude-{family}-{variant}-{YYYYMMDD}`
+
+Examples:
+- `claude-3-5-sonnet-20240620` (June 20, 2024)
+- `claude-3-7-sonnet-20250219` (Feb 19, 2025)
+
+### Likely Model ID
+
+⚠️ **Must verify from Anthropic docs!**
+
+Based on pattern, likely:
+- `claude-4-sonnet-20250929` (most likely)
+- `claude-4-5-sonnet-20250929`
+- `claude-sonnet-4-5-20250929`
+
+**Verify at**: https://docs.anthropic.com/en/docs/models-overview
+
+---
+
+## ❓ FAQ
+
+### Q: Do I need to be on my server locally?
+
+**A**: Only if:
+- Your MongoDB is local-only (no remote access)
+- You need to set environment variables on the server
+
+If MongoDB is accessible remotely, you can add models from anywhere.
+
+### Q: Will this require code changes?
+
+**A**: No! The system is configuration-driven through MongoDB.
+
+### Q: What if I use the wrong model ID?
+
+**A**: The benchmark will fail with an API error. Just update it in MongoDB:
+
+```bash
+python scripts/add_model_to_db.py --provider anthropic --model-id CORRECT_ID
+```
+
+### Q: How do I know if it worked?
+
+**A**: Run a test benchmark:
+
+```bash
+python api/bench_headless.py --providers anthropic --limit 1
+```
+
+Check for errors in the `errors_cloud` collection.
+
+### Q: Can I add multiple models at once?
+
+**A**: Yes! Run the script multiple times:
+
+```bash
+python scripts/add_model_to_db.py --provider anthropic --model-id model-1
+python scripts/add_model_to_db.py --provider anthropic --model-id model-2
+```
+
+### Q: How do I list what's already in the database?
+
+**A**: 
+
+```bash
+python scripts/add_model_to_db.py --list --provider anthropic
+```
+
+---
+
+## 📋 Prerequisites
+
+Before you start, ensure you have:
+
+### Environment Variables
+```bash
+export MONGODB_URI="mongodb+srv://..."
+export MONGODB_DB="llm-bench"
+export ANTHROPIC_API_KEY="sk-ant-..."
+```
+
+### Tools Installed
+- Python 3.9+
+- `pymongo` package: `pip install pymongo`
+- `anthropic` package: `pip install anthropic`
+- OR `mongosh` for bash scripts
+
+### Access
+- MongoDB read/write access
+- Valid Anthropic API key
+- Correct model ID from Anthropic docs
+
+---
+
+## 🎓 Step-by-Step Tutorial
+
+### Step 1: Verify Environment
+
+```bash
+# Check MongoDB connection
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval 'db.runCommand({ping: 1})'
+
+# Check Anthropic API key
+echo $ANTHROPIC_API_KEY
+```
+
+### Step 2: Find the Correct Model ID
+
+```bash
+# Test a likely model ID
+python scripts/test_anthropic_models.py --test-model claude-4-sonnet-20250929
+
+# If it fails, try variations or check Anthropic docs
+```
+
+### Step 3: Add to Database
+
+```bash
+# Add the model
+python scripts/add_model_to_db.py \
+  --provider anthropic \
+  --model-id claude-4-sonnet-20250929
+```
+
+### Step 4: Verify
+
+```bash
+# List all Anthropic models
+python scripts/add_model_to_db.py --list --provider anthropic
+```
+
+### Step 5: Test
+
+```bash
+# Run a single benchmark
+python api/bench_headless.py --providers anthropic --limit 1 --fresh-minutes 0
+```
+
+### Step 6: Check Results
+
+```bash
+# Check for errors
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.errors_cloud.find({provider: "anthropic"}).sort({ts: -1}).limit(5).pretty()
+'
+
+# Check for successful results
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.metrics_cloud_staging.find({
+  provider: "anthropic",
+  model_name: /claude-4/
+}).sort({gen_ts: -1}).limit(1).pretty()
+'
+```
+
+---
+
+## 🔗 Key Files in Codebase
+
+### Runtime
+- `api/llm_bench/models_db.py` - Loads models from MongoDB
+- `api/llm_bench/cloud/providers/anthropic.py` - Anthropic integration
+- `api/bench_headless.py` - Benchmark runner
+
+### Configuration
+- `cloud/models.json` - Reference file (NOT used at runtime)
+- Environment variables - Actual configuration
+
+### Scripts (Original)
+- `scripts/seed_model.js` - MongoDB script to add models
+- `scripts/enqueue_job.js` - Queue a benchmark job
+
+### Scripts (NEW - Created for You)
+- `scripts/add_model_to_db.py` ⭐ - Python helper
+- `scripts/test_anthropic_models.py` ⭐ - Test API & find model IDs
+- `scripts/add_sonnet_4.5.sh` - Bash helper
+
+---
+
+## 🎉 Summary
+
+### What You Learned
+
+1. ✅ Models are stored in **MongoDB** (not code)
+2. ✅ Adding models = **updating the database**
+3. ✅ The system is **configuration-driven**
+4. ✅ Changes are **immediate** (no restart needed)
+
+### What You Got
+
+1. 📚 **4 documentation files** explaining everything
+2. 🔧 **3 utility scripts** to make it easy
+3. 🎯 **Step-by-step instructions** for success
+
+### What You Need to Do
+
+1. 🔍 **Find the correct model ID** (use `test_anthropic_models.py`)
+2. ➕ **Add it to MongoDB** (use `add_model_to_db.py`)
+3. ✅ **Test it** (run a benchmark)
+4. 🎊 **Done!**
+
+---
+
+## 🚀 Ready to Start?
+
+Pick your starting point:
+
+- **Just want to do it**: [QUICK_START.md](QUICK_START.md)
+- **Want to understand first**: [RESEARCH_SUMMARY.md](RESEARCH_SUMMARY.md)
+- **Want step-by-step guide**: [ADD_SONNET_4.5_INSTRUCTIONS.md](ADD_SONNET_4.5_INSTRUCTIONS.md)
+- **Want to see what's available**: [FILES_CREATED.md](FILES_CREATED.md)
+
+Or just run:
+```bash
+python scripts/test_anthropic_models.py
+```
+
+Good luck! 🎯
\ No newline at end of file
diff --git a/RESEARCH_SUMMARY.md b/RESEARCH_SUMMARY.md
new file mode 100644
index 0000000..570b631
--- /dev/null
+++ b/RESEARCH_SUMMARY.md
@@ -0,0 +1,374 @@
+# Research Summary: Adding Models to LLM Benchmarks Database
+
+**Date**: September 30, 2025  
+**Topic**: How models are added to the database & adding Claude Sonnet 4.5
+
+---
+
+## Executive Summary
+
+✅ **Yes, this is fixable via code!** You can add new models to the database using:
+1. MongoDB scripts (recommended)
+2. Python script (created for you)
+3. Direct MongoDB commands
+
+You may need to be on your server locally only to ensure you have access to the MongoDB instance and can run the seeding scripts.
+
+---
+
+## How the System Works
+
+### Database Architecture
+
+The LLM Benchmarks system uses **MongoDB** to manage model configurations:
+
+- **Database**: `MONGODB_DB` (default: `llm-bench`)
+- **Collection**: `MONGODB_COLLECTION_MODELS` (default: `models`)
+- **Document Structure**:
+  ```javascript
+  {
+    provider: "anthropic",
+    model_id: "claude-3-5-sonnet-20240620",
+    enabled: true,
+    added_at: ISODate("2024-06-20T...")
+  }
+  ```
+
+### How Models Are Loaded
+
+The Python code (`/workspace/api/llm_bench/models_db.py`) queries MongoDB for enabled models:
+
+```python
+def load_provider_models() -> Dict[str, List[str]]:
+    # Connects to MongoDB
+    # Queries: db.models.find({"enabled": True})
+    # Returns: {provider: [model_ids]}
+```
+
+This means:
+- **Runtime**: Models are loaded from MongoDB
+- **Configuration**: `/workspace/cloud/models.json` is a reference file (not used by runtime)
+- **Control**: Enable/disable models by changing the `enabled` field in MongoDB
+
+---
+
+## Current Anthropic Models in System
+
+From `/workspace/cloud/models.json`:
+
+```json
+"anthropic": [
+    "claude-2.1",
+    "claude-3-haiku-20240307",
+    "claude-3-5-haiku-20241022",
+    "claude-3-sonnet-20240229",
+    "claude-3-5-sonnet-20240620",
+    "claude-3-7-sonnet-20250219",  // This is the newest in your system
+    "claude-3-opus-20240229"
+]
+```
+
+**Note**: Claude 3.7 Sonnet (released Feb 2025) is already in the system.
+
+---
+
+## About Claude Sonnet 4.5
+
+### Release Information
+- **Announced**: September 29, 2025
+- **Type**: Cloud API only (no local deployment)
+- **Improvements**: Coding, finance, cybersecurity, long-duration autonomous work
+
+### Model Naming Pattern
+
+Anthropic uses date-based model identifiers following the pattern:
+```
+claude-{family}-{variant}-{YYYYMMDD}
+```
+
+Examples:
+- `claude-3-5-sonnet-20240620` (June 20, 2024)
+- `claude-3-7-sonnet-20250219` (Feb 19, 2025)
+- `claude-3-5-haiku-20241022` (Oct 22, 2024)
+
+### Likely Model ID for Sonnet 4.5
+
+⚠️ **UNCONFIRMED** - Based on naming patterns, likely one of:
+- `claude-4-sonnet-20250929` (most likely)
+- `claude-4-5-sonnet-20250929`
+- `claude-sonnet-4-5-20250929`
+
+**YOU MUST VERIFY** the exact model ID from:
+- Anthropic's API documentation: https://docs.anthropic.com/en/docs/models-overview
+- Your Anthropic console/dashboard
+- Or test via their API
+
+---
+
+## How to Add Claude Sonnet 4.5
+
+### Prerequisites
+
+1. **Access to MongoDB**:
+   ```bash
+   export MONGODB_URI="mongodb+srv://user:pass@cluster.mongodb.net"
+   export MONGODB_DB="llm-bench"  # or llmbench_staging
+   ```
+
+2. **Tools installed**:
+   - Either `mongosh` (for shell script)
+   - Or `pymongo` (for Python script)
+
+3. **Anthropic API Key**:
+   ```bash
+   export ANTHROPIC_API_KEY="sk-ant-..."
+   ```
+
+4. **Model ID**: Get the exact model identifier from Anthropic
+
+---
+
+### Method 1: Using the Python Script (Recommended)
+
+I created a helper script for you at `/workspace/scripts/add_model_to_db.py`:
+
+```bash
+# First, verify the model ID from Anthropic's docs
+# Then run:
+
+python scripts/add_model_to_db.py \
+  --provider anthropic \
+  --model-id claude-4-sonnet-20250929
+
+# To list existing models:
+python scripts/add_model_to_db.py --list --provider anthropic
+```
+
+**Features**:
+- ✅ Checks for existing models
+- ✅ Confirms before updating
+- ✅ Verifies the insertion
+- ✅ Shows next steps
+- ✅ Can list all models
+
+---
+
+### Method 2: Using the Shell Script
+
+I created a bash script at `/workspace/scripts/add_sonnet_4.5.sh`:
+
+```bash
+# Edit the script to set the correct MODEL_ID
+# Then run:
+
+bash scripts/add_sonnet_4.5.sh
+```
+
+---
+
+### Method 3: Using the Original MongoDB Script
+
+The existing seed script:
+
+```bash
+PROVIDER=anthropic \
+MODEL_ID=claude-4-sonnet-20250929 \
+mongosh "$MONGODB_URI/$MONGODB_DB" scripts/seed_model.js
+```
+
+---
+
+### Method 4: Direct MongoDB Command
+
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.models.insertOne({
+  provider: "anthropic",
+  model_id: "claude-4-sonnet-20250929",
+  enabled: true,
+  added_at: new Date()
+})
+'
+```
+
+---
+
+## Testing the Integration
+
+### 1. Verify the model was added:
+
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.models.find({
+  provider: "anthropic",
+  model_id: "claude-4-sonnet-20250929"
+}).pretty()
+'
+```
+
+### 2. Run a test benchmark:
+
+```bash
+python api/bench_headless.py \
+  --providers anthropic \
+  --limit 1 \
+  --fresh-minutes 0
+```
+
+### 3. Check for errors:
+
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.errors_cloud.find({
+  provider: "anthropic",
+  model_name: /claude-4/
+}).sort({ts: -1}).limit(5).pretty()
+'
+```
+
+### 4. View results:
+
+```bash
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval '
+db.metrics_cloud_staging.find({
+  provider: "anthropic",
+  model_name: "claude-4-sonnet-20250929"
+}).sort({gen_ts: -1}).limit(1).pretty()
+'
+```
+
+---
+
+## Files I Created for You
+
+1. **`/workspace/ADD_SONNET_4.5_INSTRUCTIONS.md`**
+   - Complete step-by-step guide
+   - Troubleshooting tips
+   - Environment setup
+
+2. **`/workspace/scripts/add_model_to_db.py`**
+   - Python script to add models
+   - List existing models
+   - Interactive confirmation
+   - Error handling
+
+3. **`/workspace/scripts/add_sonnet_4.5.sh`**
+   - Bash script specifically for Sonnet 4.5
+   - Validation checks
+   - Verification steps
+
+4. **`/workspace/RESEARCH_SUMMARY.md`** (this file)
+   - Complete research findings
+   - All methods to add models
+
+---
+
+## Key Files in the Codebase
+
+### Runtime Code
+- `/workspace/api/llm_bench/models_db.py` - Loads models from MongoDB
+- `/workspace/api/llm_bench/cloud/providers/anthropic.py` - Anthropic API integration
+- `/workspace/api/bench_headless.py` - Runs benchmarks
+
+### Configuration
+- `/workspace/cloud/models.json` - Reference only (not used at runtime)
+- Environment variables - The actual configuration source
+
+### Scripts
+- `/workspace/scripts/seed_model.js` - MongoDB script to add single model
+- `/workspace/scripts/add_model_to_db.py` - **NEW**: Python helper script
+- `/workspace/scripts/add_sonnet_4.5.sh` - **NEW**: Bash helper script
+
+---
+
+## Important Notes
+
+### Do You Need Local Server Access?
+
+**Maybe** - You need access to:
+1. ✅ **MongoDB instance**: To add the model to the database
+2. ✅ **Environment variables**: To set `MONGODB_URI`, `ANTHROPIC_API_KEY`, etc.
+3. ❌ **Not needed**: Local model files (Anthropic is cloud API only)
+4. ❌ **Not needed**: Special hardware/GPU (for API-based models)
+
+**If you can**:
+- Connect to the MongoDB instance from anywhere → Can add the model remotely
+- Set environment variables on the server → Can run benchmarks
+
+### Cloud vs Local Benchmarks
+
+This system has two types:
+- **Cloud benchmarks**: Test API-based models (OpenAI, Anthropic, etc.) - This is what you want
+- **Local benchmarks**: Test locally hosted models (Llama, etc.) - Not relevant here
+
+Anthropic models are **cloud benchmarks** - they just need API access.
+
+---
+
+## Next Steps
+
+1. **Get the exact model ID**:
+   - Check https://docs.anthropic.com/en/docs/models-overview
+   - Or use Anthropic's API to list available models
+   - The date suffix is usually the release date (YYYYMMDD)
+
+2. **Ensure you have MongoDB access**:
+   ```bash
+   mongosh "$MONGODB_URI/$MONGODB_DB" --eval 'db.models.countDocuments()'
+   ```
+
+3. **Add the model** using one of the methods above
+
+4. **Test it** with a single benchmark run
+
+5. **Monitor** the first few runs for any issues
+
+---
+
+## Troubleshooting
+
+### "Model not found" errors
+- Check the model_id exactly matches what Anthropic expects
+- Verify with a direct API call first
+
+### "Authentication failed"
+- Check `ANTHROPIC_API_KEY` is set correctly
+- Verify the key has access to the new model
+
+### Model not appearing in benchmarks
+- Check `enabled: true` in database
+- Verify `fresh-minutes` isn't skipping it
+- Check the scheduler picked it up
+
+### API rate limits
+- Anthropic has rate limits per model
+- May need to adjust `SLEEP_SECONDS` between runs
+
+---
+
+## Questions to Answer
+
+Before proceeding, you should:
+
+1. ✅ **Verify the model ID**: What is the exact identifier from Anthropic?
+2. ✅ **Check API access**: Can you call this model via Anthropic's API?
+3. ✅ **Confirm MongoDB access**: Can you connect and write to the database?
+4. ✅ **Test environment**: Are all environment variables set correctly?
+
+---
+
+## Summary
+
+**The good news**: This is 100% fixable via code. The system is designed to support adding new models easily through the database.
+
+**What you need**:
+1. The correct model ID from Anthropic
+2. Access to your MongoDB instance
+3. Your Anthropic API key
+
+**What you do**:
+1. Run one of the scripts I created
+2. Test with a benchmark
+3. Monitor for any issues
+
+**No code changes needed** - everything is configuration-driven through MongoDB!
\ No newline at end of file
diff --git a/scripts/add_model_to_db.py b/scripts/add_model_to_db.py
new file mode 100755
index 0000000..55969fe
--- /dev/null
+++ b/scripts/add_model_to_db.py
@@ -0,0 +1,217 @@
+#!/usr/bin/env python3
+"""
+Script to add a model to the LLM Benchmarks MongoDB database.
+
+This provides a Python alternative to using the seed_model.js script.
+
+Usage:
+    python scripts/add_model_to_db.py --provider anthropic --model-id claude-4-sonnet-20250929
+
+Requirements:
+    - pymongo installed (pip install pymongo)
+    - MONGODB_URI environment variable set
+"""
+
+import argparse
+import os
+import sys
+from datetime import datetime
+
+try:
+    from pymongo import MongoClient
+except ImportError:
+    print("❌ Error: pymongo not installed")
+    print("   Install it with: pip install pymongo")
+    sys.exit(1)
+
+
+def add_model_to_db(provider: str, model_id: str, enabled: bool = True) -> bool:
+    """Add a model to the MongoDB models collection."""
+    
+    # Get MongoDB configuration from environment
+    uri = os.getenv("MONGODB_URI")
+    if not uri:
+        print("❌ Error: MONGODB_URI environment variable not set")
+        print("   Example: export MONGODB_URI='mongodb+srv://user:pass@cluster.mongodb.net'")
+        return False
+    
+    db_name = os.getenv("MONGODB_DB", "llm-bench")
+    coll_name = os.getenv("MONGODB_COLLECTION_MODELS", "models")
+    
+    print(f"Configuration:")
+    print(f"  Provider:   {provider}")
+    print(f"  Model ID:   {model_id}")
+    print(f"  Enabled:    {enabled}")
+    print(f"  Database:   {db_name}")
+    print(f"  Collection: {coll_name}")
+    print()
+    
+    # Create the document
+    doc = {
+        "provider": provider,
+        "model_id": model_id,
+        "enabled": enabled,
+        "added_at": datetime.now(),
+    }
+    
+    try:
+        # Connect to MongoDB
+        print("Connecting to MongoDB...")
+        client = MongoClient(uri)
+        
+        # Get collection
+        coll = client[db_name][coll_name]
+        
+        # Check if model already exists
+        existing = coll.find_one({"provider": provider, "model_id": model_id})
+        if existing:
+            print(f"⚠️  Warning: Model already exists in database")
+            print(f"   Existing document: {existing}")
+            
+            response = input("\nUpdate it? (y/N): ").strip().lower()
+            if response == 'y':
+                result = coll.update_one(
+                    {"provider": provider, "model_id": model_id},
+                    {"$set": {"enabled": enabled, "updated_at": datetime.now()}}
+                )
+                print(f"✅ Model updated successfully! (matched: {result.matched_count}, modified: {result.modified_count})")
+            else:
+                print("Skipped update.")
+                return True
+        else:
+            # Insert the document
+            print("Inserting model into database...")
+            result = coll.insert_one(doc)
+            print(f"✅ Model added successfully! ID: {result.inserted_id}")
+        
+        # Verify
+        print("\nVerifying...")
+        found = coll.find_one({"provider": provider, "model_id": model_id})
+        if found:
+            print("Document in database:")
+            for key, value in found.items():
+                if key != "_id":
+                    print(f"  {key}: {value}")
+        
+        client.close()
+        return True
+        
+    except Exception as e:
+        print(f"❌ Error: {e}")
+        return False
+
+
+def list_models(provider: str = None):
+    """List all models in the database."""
+    
+    uri = os.getenv("MONGODB_URI")
+    if not uri:
+        print("❌ Error: MONGODB_URI environment variable not set")
+        return False
+    
+    db_name = os.getenv("MONGODB_DB", "llm-bench")
+    coll_name = os.getenv("MONGODB_COLLECTION_MODELS", "models")
+    
+    try:
+        client = MongoClient(uri)
+        coll = client[db_name][coll_name]
+        
+        query = {"enabled": True}
+        if provider:
+            query["provider"] = provider
+        
+        models = list(coll.find(query, {"provider": 1, "model_id": 1, "enabled": 1, "_id": 0}))
+        
+        if not models:
+            print(f"No enabled models found" + (f" for provider: {provider}" if provider else ""))
+            return True
+        
+        print(f"\nEnabled models" + (f" for provider: {provider}" if provider else "") + ":")
+        print(f"{'Provider':<20} {'Model ID':<50} {'Enabled':<10}")
+        print("-" * 80)
+        
+        for model in sorted(models, key=lambda x: (x.get('provider', ''), x.get('model_id', ''))):
+            provider_name = model.get('provider', 'N/A')
+            model_id = model.get('model_id', 'N/A')
+            enabled = model.get('enabled', False)
+            print(f"{provider_name:<20} {model_id:<50} {enabled!s:<10}")
+        
+        print(f"\nTotal: {len(models)} models")
+        
+        client.close()
+        return True
+        
+    except Exception as e:
+        print(f"❌ Error: {e}")
+        return False
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Add or manage models in the LLM Benchmarks database",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  # Add Claude Sonnet 4.5
+  python scripts/add_model_to_db.py --provider anthropic --model-id claude-4-sonnet-20250929
+  
+  # Add a disabled model (won't be benchmarked)
+  python scripts/add_model_to_db.py --provider openai --model-id gpt-5 --disabled
+  
+  # List all enabled Anthropic models
+  python scripts/add_model_to_db.py --list --provider anthropic
+  
+  # List all enabled models
+  python scripts/add_model_to_db.py --list
+
+Environment Variables:
+  MONGODB_URI                  - MongoDB connection string (required)
+  MONGODB_DB                   - Database name (default: llm-bench)
+  MONGODB_COLLECTION_MODELS    - Collection name (default: models)
+        """
+    )
+    
+    parser.add_argument("--provider", help="Provider name (e.g., anthropic, openai)")
+    parser.add_argument("--model-id", help="Model identifier (e.g., claude-4-sonnet-20250929)")
+    parser.add_argument("--disabled", action="store_true", help="Add model as disabled")
+    parser.add_argument("--list", action="store_true", help="List models instead of adding")
+    
+    args = parser.parse_args()
+    
+    if args.list:
+        success = list_models(provider=args.provider)
+        sys.exit(0 if success else 1)
+    
+    if not args.provider or not args.model_id:
+        parser.print_help()
+        print("\n❌ Error: --provider and --model-id are required (unless using --list)")
+        sys.exit(1)
+    
+    # Special check for Claude Sonnet 4.5
+    if args.provider == "anthropic" and "claude-4" in args.model_id:
+        print("⚠️  Note: Please verify this is the correct model ID from Anthropic's documentation:")
+        print("   https://docs.anthropic.com/en/docs/models-overview")
+        print()
+    
+    enabled = not args.disabled
+    success = add_model_to_db(args.provider, args.model_id, enabled)
+    
+    if success:
+        print("\n" + "="*80)
+        print("Next Steps:")
+        print("="*80)
+        if args.provider == "anthropic":
+            print("1. Ensure ANTHROPIC_API_KEY is set in your environment")
+        print("2. Test the model with a benchmark run:")
+        print(f"   python api/bench_headless.py --providers {args.provider} --limit 1")
+        print("3. Check for any errors in the errors_cloud collection")
+        print()
+        print("To list all enabled models for this provider:")
+        print(f"   python scripts/add_model_to_db.py --list --provider {args.provider}")
+        print()
+    
+    sys.exit(0 if success else 1)
+
+
+if __name__ == "__main__":
+    main()
\ No newline at end of file
diff --git a/scripts/add_sonnet_4.5.sh b/scripts/add_sonnet_4.5.sh
new file mode 100755
index 0000000..5c9b4d2
--- /dev/null
+++ b/scripts/add_sonnet_4.5.sh
@@ -0,0 +1,117 @@
+#!/bin/bash
+
+# Script to add Claude Sonnet 4.5 to the LLM Benchmarks database
+# 
+# Prerequisites:
+# - mongosh installed
+# - MONGODB_URI and MONGODB_DB environment variables set
+#
+# Usage:
+#   1. First, find the exact model ID from Anthropic's documentation
+#   2. Update the MODEL_ID variable below with the correct value
+#   3. Ensure MONGODB_URI and MONGODB_DB are set in your environment
+#   4. Run this script: bash scripts/add_sonnet_4.5.sh
+
+set -e  # Exit on error
+
+# ====================================
+# CONFIGURATION - UPDATE THESE VALUES
+# ====================================
+
+# TODO: Replace with the actual model ID from Anthropic's docs
+# Common possibilities based on their naming pattern:
+#   - claude-4-sonnet-20250929
+#   - claude-4-5-sonnet-20250929
+#   - claude-sonnet-4-5-20250929
+MODEL_ID="claude-4-sonnet-20250929"  # <-- VERIFY AND UPDATE THIS
+
+PROVIDER="anthropic"
+ENABLED="true"
+
+# ====================================
+# VALIDATION
+# ====================================
+
+echo "=========================================="
+echo "Add Claude Sonnet 4.5 to LLM Benchmarks"
+echo "=========================================="
+echo ""
+
+# Check if mongosh is installed
+if ! command -v mongosh &> /dev/null; then
+    echo "❌ Error: mongosh is not installed"
+    echo "   Install it from: https://www.mongodb.com/docs/mongodb-shell/install/"
+    exit 1
+fi
+
+# Check if required environment variables are set
+if [ -z "$MONGODB_URI" ]; then
+    echo "❌ Error: MONGODB_URI environment variable is not set"
+    echo "   Example: export MONGODB_URI='mongodb+srv://user:pass@cluster.mongodb.net'"
+    exit 1
+fi
+
+if [ -z "$MONGODB_DB" ]; then
+    echo "⚠️  Warning: MONGODB_DB not set, using default: llm-bench"
+    MONGODB_DB="llm-bench"
+fi
+
+echo "Configuration:"
+echo "  Provider:  $PROVIDER"
+echo "  Model ID:  $MODEL_ID"
+echo "  Enabled:   $ENABLED"
+echo "  Database:  $MONGODB_DB"
+echo ""
+
+# Warn about model ID
+if [[ "$MODEL_ID" == "claude-4-sonnet-20250929" ]]; then
+    echo "⚠️  WARNING: You are using the default model ID"
+    echo "   Please verify this is correct from Anthropic's documentation:"
+    echo "   https://docs.anthropic.com/en/docs/models-overview"
+    echo ""
+    read -p "Continue anyway? (y/N): " -n 1 -r
+    echo
+    if [[ ! $REPLY =~ ^[Yy]$ ]]; then
+        echo "Aborted."
+        exit 0
+    fi
+fi
+
+# ====================================
+# ADD MODEL TO DATABASE
+# ====================================
+
+echo ""
+echo "Adding model to database..."
+
+PROVIDER=$PROVIDER MODEL_ID=$MODEL_ID ENABLED=$ENABLED \
+    mongosh "$MONGODB_URI/$MONGODB_DB" scripts/seed_model.js
+
+if [ $? -eq 0 ]; then
+    echo "✅ Model added successfully!"
+else
+    echo "❌ Failed to add model"
+    exit 1
+fi
+
+# ====================================
+# VERIFY
+# ====================================
+
+echo ""
+echo "Verifying model was added..."
+
+QUERY="db.models.find({provider: '$PROVIDER', model_id: '$MODEL_ID'}).pretty()"
+mongosh "$MONGODB_URI/$MONGODB_DB" --eval "$QUERY"
+
+echo ""
+echo "=========================================="
+echo "Next Steps:"
+echo "=========================================="
+echo "1. Ensure ANTHROPIC_API_KEY is set in your environment"
+echo "2. Test with: python api/bench_headless.py --providers anthropic --limit 1"
+echo "3. Check for errors in the errors_cloud collection"
+echo ""
+echo "To list all enabled Anthropic models:"
+echo "  mongosh \"\$MONGODB_URI/\$MONGODB_DB\" --eval 'db.models.find({provider: \"anthropic\", enabled: true}).pretty()'"
+echo ""
\ No newline at end of file
diff --git a/scripts/test_anthropic_models.py b/scripts/test_anthropic_models.py
new file mode 100755
index 0000000..4a3fdc8
--- /dev/null
+++ b/scripts/test_anthropic_models.py
@@ -0,0 +1,210 @@
+#!/usr/bin/env python3
+"""
+Test script to verify Anthropic API access and list available models.
+
+This helps you:
+1. Verify your ANTHROPIC_API_KEY is set correctly
+2. Test connectivity to Anthropic's API
+3. Discover the exact model IDs available to you
+
+Usage:
+    python scripts/test_anthropic_models.py
+    
+    # Test a specific model
+    python scripts/test_anthropic_models.py --test-model claude-4-sonnet-20250929
+
+Requirements:
+    pip install anthropic
+"""
+
+import argparse
+import os
+import sys
+from datetime import datetime
+
+try:
+    from anthropic import Anthropic
+except ImportError:
+    print("❌ Error: anthropic package not installed")
+    print("   Install it with: pip install anthropic")
+    sys.exit(1)
+
+
+def test_api_connection():
+    """Test basic connection to Anthropic API."""
+    
+    api_key = os.getenv("ANTHROPIC_API_KEY")
+    if not api_key:
+        print("❌ Error: ANTHROPIC_API_KEY environment variable not set")
+        print("   Get your API key from: https://console.anthropic.com/")
+        return False
+    
+    print(f"API Key: {api_key[:10]}...{api_key[-4:]} (length: {len(api_key)})")
+    print()
+    
+    try:
+        client = Anthropic(api_key=api_key)
+        print("✅ Successfully created Anthropic client")
+        return client
+    except Exception as e:
+        print(f"❌ Failed to create Anthropic client: {e}")
+        return None
+
+
+def test_model(client, model_id: str):
+    """Test a specific model with a simple query."""
+    
+    print(f"\n{'='*80}")
+    print(f"Testing Model: {model_id}")
+    print('='*80)
+    
+    try:
+        print("Sending test message...")
+        start_time = datetime.now()
+        
+        # Use non-streaming for simplicity
+        message = client.messages.create(
+            model=model_id,
+            max_tokens=50,
+            messages=[
+                {"role": "user", "content": "Say 'Hello, I am working!' and nothing else."}
+            ]
+        )
+        
+        end_time = datetime.now()
+        duration = (end_time - start_time).total_seconds()
+        
+        print(f"✅ Model {model_id} is working!")
+        print(f"\nResponse:")
+        print(f"  Content: {message.content[0].text if message.content else 'N/A'}")
+        print(f"  Input tokens: {message.usage.input_tokens}")
+        print(f"  Output tokens: {message.usage.output_tokens}")
+        print(f"  Duration: {duration:.2f}s")
+        print(f"  Stop reason: {message.stop_reason}")
+        
+        return True
+        
+    except Exception as e:
+        error_msg = str(e)
+        print(f"❌ Model test failed: {error_msg}")
+        
+        if "model:" in error_msg.lower() or "not found" in error_msg.lower():
+            print("\n💡 This might mean:")
+            print("   - The model ID is incorrect")
+            print("   - The model is not available to your API key")
+            print("   - The model name has changed")
+        elif "authentication" in error_msg.lower():
+            print("\n💡 Check your ANTHROPIC_API_KEY")
+        elif "rate" in error_msg.lower():
+            print("\n💡 You may have hit rate limits")
+        
+        return False
+
+
+def list_known_models():
+    """List models we know about from the codebase."""
+    
+    known_models = [
+        ("claude-2.1", "Claude 2.1", "Legacy"),
+        ("claude-3-haiku-20240307", "Claude 3 Haiku", "Fast & affordable"),
+        ("claude-3-5-haiku-20241022", "Claude 3.5 Haiku", "Latest Haiku"),
+        ("claude-3-sonnet-20240229", "Claude 3 Sonnet", "Balanced"),
+        ("claude-3-5-sonnet-20240620", "Claude 3.5 Sonnet", "Previous flagship"),
+        ("claude-3-7-sonnet-20250219", "Claude 3.7 Sonnet", "Latest in your DB"),
+        ("claude-3-opus-20240229", "Claude 3 Opus", "Most capable (expensive)"),
+    ]
+    
+    print("\n" + "="*80)
+    print("Known Anthropic Models")
+    print("="*80)
+    print(f"{'Model ID':<35} {'Name':<25} {'Notes':<20}")
+    print("-"*80)
+    
+    for model_id, name, notes in known_models:
+        print(f"{model_id:<35} {name:<25} {notes:<20}")
+    
+    print("\nFor Claude Sonnet 4.5 (released Sep 29, 2025), likely IDs:")
+    print("  - claude-4-sonnet-20250929")
+    print("  - claude-4-5-sonnet-20250929")
+    print("\nVerify at: https://docs.anthropic.com/en/docs/models-overview")
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Test Anthropic API access and available models",
+        formatter_class=argparse.RawDescriptionHelpFormatter
+    )
+    
+    parser.add_argument(
+        "--test-model",
+        help="Test a specific model ID (e.g., claude-4-sonnet-20250929)"
+    )
+    
+    parser.add_argument(
+        "--test-all-known",
+        action="store_true",
+        help="Test all known models from the codebase"
+    )
+    
+    args = parser.parse_args()
+    
+    print("="*80)
+    print("Anthropic API Model Test")
+    print("="*80)
+    print()
+    
+    # Test API connection
+    client = test_api_connection()
+    if not client:
+        sys.exit(1)
+    
+    # List known models
+    list_known_models()
+    
+    # Test specific model
+    if args.test_model:
+        success = test_model(client, args.test_model)
+        sys.exit(0 if success else 1)
+    
+    # Test all known models
+    if args.test_all_known:
+        print("\n" + "="*80)
+        print("Testing All Known Models")
+        print("="*80)
+        
+        models_to_test = [
+            "claude-3-5-sonnet-20240620",
+            "claude-3-7-sonnet-20250219",
+            "claude-3-5-haiku-20241022",
+        ]
+        
+        results = {}
+        for model_id in models_to_test:
+            print()
+            success = test_model(client, model_id)
+            results[model_id] = success
+        
+        print("\n" + "="*80)
+        print("Summary")
+        print("="*80)
+        for model_id, success in results.items():
+            status = "✅ Working" if success else "❌ Failed"
+            print(f"{status:<15} {model_id}")
+        
+        sys.exit(0)
+    
+    # Default: instructions
+    print("\n" + "="*80)
+    print("Next Steps")
+    print("="*80)
+    print("\n1. Test a specific model:")
+    print("   python scripts/test_anthropic_models.py --test-model claude-4-sonnet-20250929")
+    print("\n2. Test all known models:")
+    print("   python scripts/test_anthropic_models.py --test-all-known")
+    print("\n3. Once you find the correct model ID, add it to the database:")
+    print("   python scripts/add_model_to_db.py --provider anthropic --model-id <MODEL_ID>")
+    print()
+
+
+if __name__ == "__main__":
+    main()
\ No newline at end of file