sassoftware · bteleuca · Dec 8, 2025 · Dec 9, 2025 · Dec 9, 2025 · Dec 9, 2025
diff --git a/LLM-Definitions/_Base_Definition/README.md b/LLM-Definitions/_Base_Definition/README.md
diff --git a/LLM-Definitions/_Base_Definition/requirements.json b/LLM-Definitions/_Base_Definition/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_2_0/requirements.json b/LLM-Definitions/claude_2_0/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_2_1/requirements.json b/LLM-Definitions/claude_2_1/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_haiku_3/requirements.json b/LLM-Definitions/claude_haiku_3/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_opus_3/requirements.json b/LLM-Definitions/claude_opus_3/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_sonnet_3/requirements.json b/LLM-Definitions/claude_sonnet_3/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/claude_sonnet_3_5/requirements.json b/LLM-Definitions/claude_sonnet_3_5/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests tiktoken numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_flash_15_001/requirements.json b/LLM-Definitions/gemini_flash_15_001/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_flash_15_002/requirements.json b/LLM-Definitions/gemini_flash_15_002/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_flash_15_8b/requirements.json b/LLM-Definitions/gemini_flash_15_8b/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_flash_25/requirements.json b/LLM-Definitions/gemini_flash_25/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_flash_lite_25/requirements.json b/LLM-Definitions/gemini_flash_lite_25/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_pro_15/requirements.json b/LLM-Definitions/gemini_pro_15/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gemini_pro_25/requirements.json b/LLM-Definitions/gemini_pro_25/requirements.json
@@ -1,4 +1,8 @@
 [
+    {
+      "step": "upgrade pip before pip install",
+      "command": "pip3 -q install --upgrade pip setuptools wheel"
+    },
     {
        "step":"install common packages",
        "command":"pip3 -q install requests numpy==1.26.4"

diff --git a/LLM-Definitions/gpt_41_az_2025_01_01/Model-Card.md b/LLM-Definitions/gpt_41_az_2025_01_01/Model-Card.md
@@ -0,0 +1,130 @@
+# GPT-4.1 Azure OpenAI / Azure AI Foundry
+
+## Details
+
+The GPT-4.1 series is the latest iteration of the GPT-4o model family. This iteration of models is specifically targeted for better coding and instruction following, making it better at handling complex technical and coding problems.
+
+**Direct from Azure models** - GPT-4.1 is a select portfolio model curated for its market-differentiated capabilities:
+- **Secure and managed by Microsoft**: Purchase and manage models directly through Azure with a single license, consistent support, and no third-party dependencies, backed by Azure's enterprise-grade infrastructure.
+- **Streamlined operations**: Benefit from unified billing, governance, and seamless PTU portability across models hosted on Azure - all part of Microsoft Foundry.
+- **Future-ready flexibility**: Access the latest models as they become available, and easily test, deploy, or switch between them within Microsoft Foundry; reducing integration effort.
+- **Cost control and optimization**: Scale on demand with pay-as-you-go flexibility or reserve PTUs for predictable performance and savings.
+
+## Key Capabilities
+
+- **Text and image processing**: Multimodal input support
+- **JSON Mode**: Structured JSON output generation
+- **Parallel function calling**: Execute multiple function calls simultaneously
+- **Enhanced accuracy and responsiveness**: Parity with English text and coding tasks compared to GPT-4 Turbo with Vision
+- **Superior multilingual performance**: Improved performance in non-English languages and vision tasks
+- **Complex structured outputs**: Support for sophisticated output formatting
+
+## Context and Output
+
+GPT-4.1 increases the context token limit up to **1M input tokens** with separate billing for:
+- Small context: 128k tokens
+- Large context: up to 1M tokens
+
+As with the previous GPT-4o model family, it supports a **16k output size**.
+
+## Model ID
+
+The GPT-4.1 model is available through Azure OpenAI and Azure AI Foundry.
+
+**Availability**: Standard, Global Standard, Global Batch, Regional Provisioned Throughput, Global Provisioned Throughput, Data Zone Standard, Data Zone Provisioned Throughput, Data Zone Batch
+
+**Lifecycle**: Generally available (Preview)
+
+**Training cut-off date**: Not supplied by provider
+
+## Data, Media and Languages
+
+**Supported data types:**
+- **Inputs**: Text, image
+- **Outputs**: Text (up to 16k tokens)
+
+**Input Formats**: Text, image processing
+
+**Output Formats**: Text with JSON Mode and support for complex structured outputs
+
+**Supported languages**: Superior performance in non-English languages and vision tasks. Includes support for: en, it, af, es, de, fr, id, ru, pl, uk, el, lv, zh, ar, tr, ja, sw, cy, ko, is, bn, ur, ne, th, pa, mr, te, and many more.
+
+## Use Cases
+
+### Key Use Cases
+
+This iteration of models is specifically targeted for:
+- **Complex coding problems**: Better at handling technical programming challenges
+- **Instruction following**: Enhanced accuracy in following detailed instructions
+- **Technical documentation**: Processing and generating complex technical content
+- **Function calling**: Build applications that fetch data or take actions with external systems
+- **Long-context processing**: Handle large codebases or conversation histories (up to 1M tokens)
+
+### Out of Scope Use Cases
+
+Prompts and completions are passed through a default configuration of Azure AI Content Safety classification models to detect and prevent the output of harmful content. Learn more about [Azure AI Content Safety](https://learn.microsoft.com/en-us/azure/ai-services/content-safety/). Additional classification models and configuration options are available when you deploy an Azure OpenAI model in production.
+
+## Pricing
+
+Pricing is based on a number of factors, including deployment type and tokens used:
+- **Small context** (up to 128k tokens): Standard pricing
+- **Large context** (128k - 1M tokens): Separate billing for extended context
+
+See [Azure OpenAI pricing details](https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/) for current rates.
+
+## Transparency
+
+### Model Provider
+
+This model is provided through the Azure OpenAI service and Azure AI Foundry.
+
+### Distribution Channels
+
+- Azure OpenAI Service
+- Azure AI Foundry
+
+### Relevant Documents
+
+The following documents are applicable:
+
+- [Overview of Responsible AI practices for Azure OpenAI models](https://learn.microsoft.com/en-us/legal/cognitive-services/openai/overview)
+- [Transparency Note for Azure OpenAI Service](https://learn.microsoft.com/en-us/legal/cognitive-services/openai/transparency-note)
+- [Introducing the GPT-4.1 Series](https://techcommunity.microsoft.com/blog/azure-ai-services-blog/introducing-gpt-4-1-openais-new-flagship-multimodal-model-now-in-preview-on-azu/4357395): OpenAI's new flagship multimodal model now in preview on Azure
+
+### Model Architecture
+
+The provider has not supplied detailed architecture information.
+
+## Responsible AI Considerations
+
+### Built-in Safety Measures
+
+Safety is built into GPT-4.1 from the beginning, and reinforced at every step of the development process:
+
+- **Pre-training filtering**: Content filtering to exclude hate speech, adult content, personal information aggregation, and spam
+- **Post-training alignment**: Reinforcement learning with human feedback (RLHF) to improve accuracy and reliability
+- **Expert evaluation**: Assessed using both automated and human evaluations according to Azure's Preparedness Framework
+- **Instruction hierarchy**: Enhanced ability to resist jailbreaks, prompt injections, and system prompt extractions
+- **Continuous monitoring**: Ongoing safety improvements as new risks are identified
+
+### Content Filtering
+
+Prompts and completions are passed through a default configuration of Azure AI Content Safety classification models to detect and prevent the output of harmful content. Learn more about Azure AI Content Safety. Additional classification models and configuration options are available when you deploy an Azure OpenAI model in production; learn more.
+
+## Additional Information
+
+### Training and Testing
+
+The provider has not supplied detailed training, testing, and validation information.
+
+### Performance
+
+GPT-4.1 is specifically optimized for:
+- **Enhanced coding capabilities**: Better handling of complex technical and coding problems
+- **Instruction following**: Improved accuracy in following detailed instructions
+- **Parity with GPT-4 Turbo with Vision**: Equivalent performance on English text and coding tasks
+- **Superior multilingual performance**: Enhanced performance in non-English languages and vision tasks
+
+### Learn More
+
+For the latest information and updates on GPT-4.1, visit the [Azure AI Foundry Model Catalog](https://ai.azure.com).
diff --git a/LLM-Definitions/gpt_41_az_2025_01_01/README.md b/LLM-Definitions/gpt_41_az_2025_01_01/README.md
@@ -0,0 +1,140 @@
+# GPT-4.1 from Azure OpenAI / Azure AI Foundry
+
+Source: https://ai.azure.com/explore/models/gpt-4.1/version/2025-04-14/registry/azure-openai
+
+## Required Items
+
+Azure OpenAI and Azure AI Foundry provide REST APIs for interaction and response generation.
+
+To use a GPT-4.1 model, you need:
+
+- A resource (Azure OpenAI or Azure AI Foundry). See [Create and deploy an Azure OpenAI resource](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/create-resource?pivots=web-portal).
+- A deployment of the GPT-4.1 model. See below.
+- The endpoint and API key.
+
+## Create a Model Deployment
+
+Before using the Azure OpenAI or Azure AI Foundry resource, deploy the GPT-4.1 model.
+
+### Option A: Deploy via Azure AI Foundry
+
+1. Navigate to [Azure AI Foundry](https://ai.azure.com).
+1. Select your project or create a new one.
+1. **Create a deployment**:
+   * Search for and select **GPT-4.1**.
+   * Deployment name: **gpt-4.1** (**MUST USE THIS EXACT NAME**).
+   * **Deployment type**: Choose based on your needs (e.g., Global Standard).
+   * **Model version**: Choose the latest available.
+   * Tokens per Minute Rate Limit: Choose around 250K, if possible.
+   * Click **Deploy**.
+
+### Option B: Deploy via Azure OpenAI Studio
+
+1. Inside your Azure OpenAI resource, click **Go to Azure OpenAI Studio** or navigate directly to [oai.azure.com](https://oai.azure.com).
+1. Go to **Deployments** > **Create new deployment**.
+1. Select **GPT-4.1**:
+   * Deployment name: **gpt-4.1** (**MUST USE THIS EXACT NAME**).
+   * **Model version**: Choose the latest available.
+   * Location: Choose your preferred region.
+   * Click **Deploy**.
+
+The chat playground will open. Feel free to ask a question to test the deployment.
+
+## Retrieve your Azure OpenAI or Azure AI Foundry Key and Endpoint
+
+Steps:
+
+1. On the Azure portal:
+   * Search for **Azure OpenAI** or **Azure AI Foundry**.
+1. Locate your service.
+1. Expand **Resource Management** > **Keys and Endpoint**.
+    - **Endpoint**: Found in the **Keys & Endpoint** section of the Azure portal.
+      - Azure OpenAI example: `https://westus3.api.cognitive.microsoft.com/`
+      - Azure AI Foundry example: `https://sbxbotres.cognitiveservices.azure.com/`
+    - **Note**: Save the hostname (e.g., `westus3.api.cognitive.microsoft.com` or `sbxbotres.cognitiveservices.azure.com`). You'll configure this as the `azure_openai_resource` option when scoring the model.
+    - **API Key**: Found in the **Keys & Endpoint** section. Copy any of the keys. You will need it when scoring the model (passed as `API_KEY` in options).
+
+## Configuration Options
+
+This model is **location-independent** and supports both Azure OpenAI and Azure AI Foundry deployments. Configure your endpoint at runtime through the `options` parameter - no code changes required for different Azure regions or deployment types.
+
+### Required Options:
+
+- **`API_KEY`**: Your API key from the Azure portal (required for authentication)
+
+### Endpoint Configuration (choose one):
+
+**Option 1: Full Endpoint URL Override** (recommended for Azure AI Foundry)
+- **`endpoint_url`**: Complete chat completions endpoint URL
+  - Example (Azure AI Foundry): `https://sbxbotres.cognitiveservices.azure.com/openai/deployments/gpt-4.1/chat/completions?api-version=2025-01-01-preview`
+  - Example (Azure OpenAI): `https://westus3.api.cognitive.microsoft.com/openai/deployments/gpt-4.1/chat/completions?api-version=2025-01-01-preview`
+  - When set, this is used directly (ignores `azure_openai_resource` and `api_version`)
+
+**Option 2: Resource Hostname** (constructed endpoint)
+- **`azure_openai_resource`**: Your Azure resource hostname
+  - Azure AI Foundry: `sbxbotres.cognitiveservices.azure.com`
+  - Azure OpenAI: `westus3.api.cognitive.microsoft.com` or short name `my-openai-westus3` (auto-appends `.openai.azure.com`)
+  - Find this in the **Keys & Endpoint** section of your resource
+- **`api_version`**: Azure OpenAI API version (default: `2025-01-01-preview`)
+  - Used only when `endpoint_url` is not provided
+
+### Optional Parameters:
+
+- **`temperature`**: Sampling temperature 0-2 (default: `1`)
+  - Higher values (0.8) make output more random
+  - Lower values (0.2) make output more focused and deterministic
+- **`top_p`**: Nucleus sampling 0-1 (default: `1`)
+  - Controls diversity via nucleus sampling
+  - 0.1 means only tokens in top 10% probability are considered
+
+### Example Options Strings:
+
+**Azure AI Foundry with full endpoint URL:**
+```
+{endpoint_url:https://sbxbotres.cognitiveservices.azure.com/openai/deployments/gpt-4.1/chat/completions?api-version=2025-01-01-preview,temperature:1,top_p:1,API_KEY:your-key-here}
+```
+
+**Azure OpenAI with resource hostname:**
+```
+{azure_openai_resource:westus3.api.cognitive.microsoft.com,api_version:2025-01-01-preview,temperature:1,top_p:1,API_KEY:your-key-here}
+```
+
+**Azure OpenAI with short resource name:**
+```
+{azure_openai_resource:my-openai-westus3,api_version:2025-01-01-preview,temperature:1,top_p:1,API_KEY:your-key-here}
+```
+
+### Testing Locally:
+
+To test the model wrapper locally, uncomment the example code at the bottom of `gpt41MiniScore.py`:
+
+**Using Azure AI Foundry:**
+```python
+if __name__ == "__main__":
+    userPrompt = ["Count to ten in French"]
+    systemPrompt = ["You are an AI Assistant helping people learn languages"]
+    options = ["{endpoint_url:https://sbxbotres.cognitiveservices.azure.com/openai/deployments/gpt-4.1/chat/completions?api-version=2025-01-01-preview,temperature:1,top_p:1,API_KEY:your-actual-key}"]
+    response, run_time, prompt_length, output_length = scoreModel(userPrompt, systemPrompt, options)
+```
+
+**Using Azure OpenAI:**
+```python
+if __name__ == "__main__":
+    userPrompt = ["Count to ten in French"]
+    systemPrompt = ["You are an AI Assistant helping people learn languages"]
+    options = ["{azure_openai_resource:westus3.api.cognitive.microsoft.com,api_version:2025-01-01-preview,temperature:1,top_p:1,API_KEY:your-actual-key}"]
+    response, run_time, prompt_length, output_length = scoreModel(userPrompt, systemPrompt, options)
+```
+
+Run: `python gpt41MiniScore.py`
+
+## Benefits of This Approach
+
+- **Dual deployment support**: Works with both Azure OpenAI and Azure AI Foundry
+- **Multi-region support**: Works with any Azure region without code changes
+- **Flexible configuration**: Choose between full endpoint URL or constructed endpoint
+- **Configurable at runtime**: Endpoint and API version configured via options
+- **Development flexibility**: Different endpoints for dev/test/prod environments
+- **Future-proof**: Easy to update API versions as Azure releases new features
+
+## End