Docs/nebius quickstart

README.md

@@ -23,6 +23,14 @@ As mentioned above, this is a research demonstration prototype and should not be
 
 ## Getting Started
 
+### Nebius AI Studio quickstart (OpenAI-compatible)
+
+- Set NEBIUS_API_KEY in backend/.env and choose DEFAULT_MODEL=nebius-kimi-k2.
+- Use FRAME_BASE_URL=https://api.studio.nebius.com/v1/ and FRAME_MODEL=moonshotai/Kimi-K2-Instruct.
+- Frontend: NEXT_PUBLIC_API_VERSION=v1, NEXT_PUBLIC_ENABLE_V2_API=false, DRY_RUN=true (no Tavily) or false (with Tavily).
+- See Nebius AI Studio Cookbook for more examples: https://github.com/nebius/ai-studio-cookbook
+
+
 To run the prototype, you need to start both the backend and frontend services:
 
 ### 1. Backend Setup

backend/README.md

@@ -404,3 +404,31 @@ For issues and questions:
 - Create an issue in the repository
 - Check the logs in the `logs/` directory
 - Review the configuration settings
+
+### Nebius AI Studio (OpenAI-compatible)
+
+To run the v1 endpoint with Nebius as the default LLM:
+
+1) Copy env and set CORS to your frontend port (e.g., 3004):
+
+```bash
+cp env.example .env
+sed -i '' 's#^FRONTEND_URL=.*#FRONTEND_URL=http://localhost:3004#' .env
+```
+
+2) Set Nebius defaults and provide the key (no quotes):
+
+```bash
+sed -i '' 's#^DEFAULT_MODEL=.*#DEFAULT_MODEL=nebius-kimi-k2#' .env
+# NEBIUS_API_KEY=your-nebius-api-key
+# FRAME_BASE_URL=https://api.studio.nebius.com/v1/
+# FRAME_MODEL=moonshotai/Kimi-K2-Instruct
+```
+
+3) Start:
+
+```bash
+./launch_server.sh
+```
+
+Notes: Tavily search optional via tavily_api.txt.

backend/env.example

@@ -1,47 +1,98 @@
 # Universal Deep Research Backend (UDR-B) - Environment Configuration
+# Universal Deep Research Backend (UDR-B) - Environment Configuration
+# Copy this file to .env and customize the values for your deployment
 # Copy this file to .env and customize the values for your deployment
 
+
 # Server Configuration
+# Server Configuration
+HOST=0.0.0.0
 HOST=0.0.0.0
 PORT=8000
+PORT=8000
 LOG_LEVEL=info
+LOG_LEVEL=info
+
 
+# CORS Configuration
 # CORS Configuration
 FRONTEND_URL=http://localhost:3000
+FRONTEND_URL=http://localhost:3000
+ALLOW_CREDENTIALS=true
 ALLOW_CREDENTIALS=true
 
+
 # Model Configuration
+# Model Configuration
+DEFAULT_MODEL=llama-3.1-nemotron-253b
 DEFAULT_MODEL=llama-3.1-nemotron-253b
 LLM_BASE_URL=https://integrate.api.nvidia.com/v1
+LLM_BASE_URL=https://integrate.api.nvidia.com/v1
 LLM_API_KEY_FILE=nvdev_api.txt
+LLM_API_KEY_FILE=nvdev_api.txt
+LLM_TEMPERATURE=0.2
 LLM_TEMPERATURE=0.2
 LLM_TOP_P=0.7
+LLM_TOP_P=0.7
 LLM_MAX_TOKENS=2048
+LLM_MAX_TOKENS=2048
+
 
+# Search Configuration
 # Search Configuration
 TAVILY_API_KEY_FILE=tavily_api.txt
+TAVILY_API_KEY_FILE=tavily_api.txt
+MAX_SEARCH_RESULTS=10
 MAX_SEARCH_RESULTS=10
 
+
+# Research Configuration
 # Research Configuration
 MAX_TOPICS=1
+MAX_TOPICS=1
 MAX_SEARCH_PHRASES=1
+MAX_SEARCH_PHRASES=1
+MOCK_DIRECTORY=mock_instances/stocks_24th_3_sections
 MOCK_DIRECTORY=mock_instances/stocks_24th_3_sections
 RANDOM_SEED=42
+RANDOM_SEED=42
 
+
+# Logging Configuration
 # Logging Configuration
 LOG_DIR=logs
+LOG_DIR=logs
 TRACE_ENABLED=true
+TRACE_ENABLED=true
+COPY_INTO_STDOUT=false
 COPY_INTO_STDOUT=false
 
+
 # FrameV4 Configuration
+# FrameV4 Configuration
+LONG_CONTEXT_CUTOFF=8192
 LONG_CONTEXT_CUTOFF=8192
 FORCE_LONG_CONTEXT=false
+FORCE_LONG_CONTEXT=false
 MAX_ITERATIONS=1024
+MAX_ITERATIONS=1024
+INTERACTION_LEVEL=none
 INTERACTION_LEVEL=none
 
+
 # Model-specific overrides (optional)
+# Model-specific overrides (optional)
+# LLAMA_3_1_8B_BASE_URL=https://integrate.api.nvidia.com/v1
 # LLAMA_3_1_8B_BASE_URL=https://integrate.api.nvidia.com/v1
 # LLAMA_3_1_8B_MODEL=nvdev/meta/llama-3.1-8b-instruct
+# LLAMA_3_1_8B_MODEL=nvdev/meta/llama-3.1-8b-instruct
 # LLAMA_3_1_8B_TEMPERATURE=0.2
+# LLAMA_3_1_8B_TEMPERATURE=0.2
+# LLAMA_3_1_8B_TOP_P=0.7
 # LLAMA_3_1_8B_TOP_P=0.7
-# LLAMA_3_1_8B_MAX_TOKENS=2048 
+# LLAMA_3_1_8B_MAX_TOKENS=2048 # LLAMA_3_1_8B_MAX_TOKENS=2048 
+# Nebius AI Studio (OpenAI-compatible) configuration (uncomment and set to enable)
+# DEFAULT_MODEL=nebius-kimi-k2
+# FRAME_BASE_URL=https://api.studio.nebius.com/v1/
+# FRAME_MODEL=moonshotai/Kimi-K2-Instruct
+# NEBIUS_API_KEY=your-nebius-api-key

frontend/README.md

@@ -169,3 +169,17 @@ The application can be deployed to any platform that supports Next.js:
 This software is provided for research and demonstration purposes only. Please refer to the [DISCLAIMER](DISCLAIMER.txt) file for complete terms and conditions regarding the use of this software. You can find the license in [LICENSE](LICENSE.txt).
 
 **Do not use this code in production.**
+
+### Using Nebius AI Studio (OpenAI-compatible)
+
+Local dev with Nebius v1:
+
+```bash
+cp env.example .env.local
+sed -i '' 's#^NEXT_PUBLIC_API_VERSION=.*#NEXT_PUBLIC_API_VERSION=v1#' .env.local
+sed -i '' 's#^NEXT_PUBLIC_ENABLE_V2_API=.*#NEXT_PUBLIC_ENABLE_V2_API=false#' .env.local
+sed -i '' 's#^NEXT_PUBLIC_DRY_RUN=.*#NEXT_PUBLIC_DRY_RUN=false#' .env.local
+npm run dev -- -p 3004
+```
+
+Backend must be configured with `DEFAULT_MODEL=nebius-kimi-k2` and `NEBIUS_API_KEY` in `backend/.env`.