{"id":31644564,"url":"https://github.com/jtgsystems/universal-ollama-optimizer","last_synced_at":"2026-05-06T03:35:12.047Z","repository":{"id":315006813,"uuid":"1057702218","full_name":"jtgsystems/universal-ollama-optimizer","owner":"jtgsystems","description":"Professional bash script for launching and managing Ollama AI models with parameter profile suggestions, system monitoring, and automated configuration. Not an automatic optimizer - provides suggested /set commands for manual application.","archived":false,"fork":false,"pushed_at":"2026-02-09T17:43:48.000Z","size":736,"stargazers_count":0,"open_issues_count":1,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-09T21:12:45.375Z","etag":null,"topics":["ai","ai-models","automation","bash","launcher","linux","llm","local-ai","model-management","model-optimizer","ollama","optimization","shell","terminal-ui","tools"],"latest_commit_sha":null,"homepage":null,"language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jtgsystems.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-09-16T05:15:54.000Z","updated_at":"2026-02-05T03:09:02.000Z","dependencies_parsed_at":"2025-09-16T07:15:29.177Z","dependency_job_id":"c2f30499-fffc-40e1-8067-f2da18c27c1b","html_url":"https://github.com/jtgsystems/universal-ollama-optimizer","commit_stats":null,"previous_names":["jtgsystems/universal-ollama-optimizer"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/jtgsystems/universal-ollama-optimizer","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jtgsystems%2Funiversal-ollama-optimizer","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jtgsystems%2Funiversal-ollama-optimizer/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jtgsystems%2Funiversal-ollama-optimizer/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jtgsystems%2Funiversal-ollama-optimizer/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jtgsystems","download_url":"https://codeload.github.com/jtgsystems/universal-ollama-optimizer/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jtgsystems%2Funiversal-ollama-optimizer/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32677932,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-06T02:33:58.958Z","status":"ssl_error","status_checked_at":"2026-05-06T02:33:39.611Z","response_time":117,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","ai-models","automation","bash","launcher","linux","llm","local-ai","model-management","model-optimizer","ollama","optimization","shell","terminal-ui","tools"],"created_at":"2025-10-07T04:54:01.178Z","updated_at":"2026-05-06T03:35:12.031Z","avatar_url":"https://github.com/jtgsystems.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"![Banner](banner.png)\n\n#  Ollama Launcher \u0026 Manager\n\n**Developed by [JTGSYSTEMS.COM](https://jtgsystems.com) | [JointTechnologyGroup.com](https://jointtechnologygroup.com)**\n\nA professional bash script for launching and managing Ollama AI models with intelligent profile suggestions, system monitoring, and automated configuration.\n\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)\n[![Platform](https://img.shields.io/badge/platform-Linux-green.svg)](https://github.com/jtgsystems/universal-ollama-optimizer)\n[![Ollama](https://img.shields.io/badge/ollama-compatible-orange.svg)](https://ollama.ai)\n[![Bash](https://img.shields.io/badge/bash-5.0+-blue.svg)](https://www.gnu.org/software/bash/)\n\n##  What This Actually Does\n\nThis is a **model launcher and management toolkit** for Ollama. It provides:\n\n- **Interactive model selection** from 55+ pre-configured 2025 models\n- **System validation** before launching (network, Ollama service, disk space, GPU/RAM checks)\n- **Parameter profile suggestions** with recommended settings for different use cases\n- **Automatic model downloading** with validation and retry logic\n- **Clean terminal UI** with colors and progress indicators\n\n### Important Note on \"Optimization\"\n\nThis script **does NOT automatically apply parameters** to models. Ollama's CLI doesn't support passing parameters like `temperature` or `top_p` at launch time. Instead, this tool:\n\n1. **Suggests optimized profiles** based on your use case\n2. **Shows you the recommended parameters** for that profile\n3. **Provides copy-paste ready `/set` commands** to apply manually during your session\n\nFor automatic parameter application, you would need to use Ollama's HTTP API directly or create custom Modelfiles (which this script can help with).\n\n##  Latest Updates (September 2025)\n\n**Added 5 Critical Missing Models Based on Community Research:**\n-  **GLM-4** - Ranks 3rd overall on hardcore benchmarks, beats Llama 3 8B\n-  **Magicoder** - OSS-Instruct trained coding specialist with 75K synthetic instruction data\n-  **Gemma3n** - Multimodal model optimized for everyday devices (phones, tablets, laptops)\n-  **Granite 3.3** - IBM's improved models with 128K context (2B and 8B variants)\n-  **Gemma3:270M** - Ultra-compact 270M model with 0.75% battery usage on mobile devices\n\n**Enhanced Menu Organization:**\n- Expanded model selection from 50+ to 55+ latest 2025 models\n- Added new performance benchmarks and descriptions\n- Updated all optimization profiles and system recommendations\n\n##  Features\n\n- ** Universal Model Support** - Works with any Ollama model (llama3.2, mistral, gemma2, etc.)\n- ** 6 Profile Suggestions** - Pre-configured parameter suggestions: Balanced, Technical, Creative, Code, Reasoning, Roleplay\n- ** System Resource Monitoring** - Real-time GPU, RAM, and disk space monitoring\n- ** Auto-Download \u0026 Validation** - Intelligent model downloading with space checking\n- ** Professional Interface** - Clean, colorful terminal UI with progress indicators\n- ** Configuration Management** - File-based config with runtime overrides\n- ** Custom Parameters** - Full manual parameter control and Modelfile creation\n- ** Error Logging** - Comprehensive logging to `~/.config/universal-ollama-optimizer/errors.log`\n\n##  Requirements\n\n- **Linux** (Ubuntu 20.04+, other distros compatible)\n- **Bash 5.0+** (pre-installed on most systems)\n- **Ollama** ([Download here](https://ollama.ai/download))\n- **8GB+ RAM** recommended\n- **GPU** optional but recommended (NVIDIA with CUDA support)\n\n##  Quick Start\n\n### One-Command Install \u0026 Run\n```bash\n# Download and run\ncurl -fsSL https://raw.githubusercontent.com/jtgsystems/universal-ollama-optimizer/main/universal-ollama-optimizer.sh -o universal-ollama-optimizer.sh\nchmod +x universal-ollama-optimizer.sh\n./universal-ollama-optimizer.sh\n```\n\n### Manual Installation\n```bash\n# Clone repository\ngit clone https://github.com/jtgsystems/universal-ollama-optimizer.git\ncd universal-ollama-optimizer\n\n# Make executable\nchmod +x universal-ollama-optimizer.sh\n\n# Run\n./universal-ollama-optimizer.sh\n```\n\n##  Usage\n\n### Interactive Mode (Default)\n```bash\n./universal-ollama-optimizer.sh\n```\n\n1. **Select Model** - Choose from available models or enter new model name\n2. **Choose Profile** - Select parameter suggestion profile (1-9)\n3. **Launch Model** - Model starts with parameter instructions displayed\n4. **Apply Parameters** - Use `/set` commands manually during the session\n\n### Example Session\n```\nEnter model name: llama3.2:latest\n\nModel Information:\n  • Model: llama3.2:latest\n  • Size: 4.7GB\n  • Parameters: 3.2B\n\nSystem Status:\n  • GPU Memory: 16380 MB\n  • System RAM: 32GB\n  • Available Disk: 150GB\n\nParameter Profiles:\n  1) Balanced          - General purpose (temp: 0.5)\n  2) Technical/Factual - Precise answers (temp: 0.2)\n  3) Creative Writing  - Imaginative content (temp: 1.0)\n  4) Code Generation   - Programming tasks (temp: 0.15)\n  5) Reasoning/Logic   - Problem solving (temp: 0.3)\n  6) Roleplay/Chat     - Conversational (temp: 0.8)\n\nSelect profile [1-6]: 1\n\nStarting llama3.2:latest with Balanced profile suggestion...\n\nTo apply these parameters during your session, use:\n/set parameter temperature 0.5\n/set parameter top_p 0.85\n/set parameter top_k 30\n/set parameter repeat_penalty 1.08\n```\n\n##  Configuration\n\n### Config File Location\n```\n~/.config/universal-ollama-optimizer/config.conf\n```\n\n### Example Configuration\n```ini\n# Default profile (balanced, technical, creative, code, reasoning, roleplay)\nDEFAULT_PROFILE=\"balanced\"\n\n# Auto-start Ollama service if not running\nAUTO_START_OLLAMA=true\n\n# System monitoring\nSHOW_SYSTEM_INFO=true\nSHOW_GPU_INFO=true\n\n# Download settings\nDOWNLOAD_TIMEOUT=1800\nMIN_DISK_SPACE_GB=5\n```\n\n##  Parameter Profiles (Suggestions)\n\n| Profile | Temperature | Top-P | Top-K | Best For |\n|---------|-------------|-------|-------|----------|\n| **Balanced** | 0.5 | 0.85 | 30 | General use, Q\u0026A |\n| **Technical** | 0.2 | 0.8 | 20 | Documentation, facts |\n| **Creative** | 1.0 | 0.95 | 50 | Stories, brainstorming |\n| **Code** | 0.15 | 0.7 | 15 | Programming, debugging |\n| **Reasoning** | 0.3 | 0.75 | 25 | Logic, analysis |\n| **Roleplay** | 0.8 | 0.9 | 40 | Character chat |\n\n##  Advanced Features\n\n### Custom Parameters\nChoose option `7` for manual parameter configuration:\n- Temperature (0.0-2.0)\n- Top-P (0.0-1.0)\n- Top-K (1-100)\n- Context length\n- Max tokens per response\n- Repeat penalty\n\n### Modelfile Creation\nChoose option `8` to create and save custom model configurations:\n```bash\n# Creates permanent optimized model variants\n# Example: llama3.2-coding, mistral-creative\n```\n\n### Runtime Commands\nOnce model is running, use these commands:\n```bash\n/set parameter temperature 0.7    # Adjust parameters\n/set system \"You are a coding assistant\"  # Change system prompt\n/show parameters                  # View current settings\n/save my-conversation            # Save session\n/load my-conversation            # Load session\n```\n\n##  Troubleshooting\n\n### Common Issues\n\n**Ollama Not Found**\n```bash\n# Install Ollama\ncurl -fsSL https://ollama.ai/install.sh | sh\n```\n\n**Permission Denied**\n```bash\n# Make script executable\nchmod +x universal-ollama-optimizer.sh\n```\n\n**Model Download Fails**\n```bash\n# Check internet connection and disk space\ndf -h\nping ollama.ai\n```\n\n**GPU Not Detected**\n```bash\n# Check NVIDIA drivers\nnvidia-smi\n```\n\n##  Project Structure\n\n```\nuniversal-ollama-optimizer/\n├── universal-ollama-optimizer.sh    # Main script\n├── README.md                        # This file\n├── LICENSE                          # MIT License\n└── .github/\n    ├── workflows/\n    │   └── test.yml                 # CI/CD tests\n    ├── ISSUE_TEMPLATE/              # Issue templates\n    └── PULL_REQUEST_TEMPLATE.md     # PR template\n```\n\n##  Contributing\n\nWe welcome contributions! Please see our [Contributing Guidelines](CONTRIBUTING.md).\n\n### Development Setup\n```bash\n# Fork and clone\ngit clone https://github.com/jtgsystems/universal-ollama-optimizer.git\ncd universal-ollama-optimizer\n\n# Test the script\n./universal-ollama-optimizer.sh\n\n# Run with debug mode\nbash -x universal-ollama-optimizer.sh\n```\n\n### Reporting Issues\n- Use the [issue tracker](https://github.com/jtgsystems/universal-ollama-optimizer/issues)\n- Include OS version, Ollama version, and error messages\n- Provide steps to reproduce\n\n##  License\n\nThis project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.\n\n##  Acknowledgments\n\n- **Jesus Christ** - Our Lord and Saviour, for all gifts and abilities\n- **Ollama Team** - For creating an excellent local AI platform\n- **Community Contributors** - For feedback and improvements\n- **JTGSYSTEMS.COM** - For development and maintenance\n\n##  Support\n\n- **Issues**: [GitHub Issues](https://github.com/jtgsystems/universal-ollama-optimizer/issues)\n- **Discussions**: [GitHub Discussions](https://github.com/jtgsystems/universal-ollama-optimizer/discussions)\n- **Website**: [JTGSYSTEMS.COM](https://jtgsystems.com)\n\n---\n\n##  Recommended Ollama Models (September 2025)\n\n*Based on latest community recommendations and performance benchmarks*\n\n### ** Top-Tier Models (September 2025)**\n\n#### ** Best Overall Performance**\n- **`llama3.3:70b`** - Meta's flagship 2025 model, rivals GPT-4 performance locally\n- **`glm4:latest`**  - Ranks 3rd overall, beats Llama 3 8B in benchmarks\n- **`llama3.1:8b`** - Community favorite, best balance of performance and efficiency\n- **`llama3.1:70b`** - High-performance for complex reasoning and enterprise use\n- **`deepseek-r1`** - Powerhouse for deep logical reasoning and analysis\n\n#### ** Premier Coding Models**\n- **`deepseek-coder:33b`** - #1 coding model, excels at complex programming tasks\n- **`magicoder:latest`**  - OSS-Instruct trained specialist, 75K synthetic data\n- **`qwen3-coder:30b`** - Alibaba's latest 2025 coding model with major improvements\n- **`codellama:34b`** - Meta's specialized coding model with excellent context understanding\n- **`qwen2.5-coder:32b`** - Previous Alibaba model with solid code generation\n\n#### ** Resource-Efficient Champions**\n- **`granite3.3:8b`**  - IBM's improved model with 128K context, rivals Llama 3.1\n- **`mistral:7b-instruct`** - Community-recommended for beginners, excellent performance/resource ratio\n- **`phi4:14b`** - Microsoft's 2025 state-of-the-art efficiency model\n- **`granite3.3:2b`**  - IBM's efficient enterprise model for edge deployment\n- **`llama3.2:3b`** - Compact Llama for lightweight deployments\n- **`gemma3:270m`**  - Ultra-compact 270M model, 0.75% battery usage on mobile\n- **`gemma2:9b`** - Google's efficient model, great for general tasks\n\n#### ** Creative \u0026 Multimodal**\n- **`gemma3n:latest`**  - Multimodal optimized for everyday devices (phones, tablets)\n- **`llava:latest`** - Leading vision model for image analysis and VQA\n- **`qwen-vl`** - Advanced multimodal model for document and image processing\n- **`gemma2:27b`** - Excellent for creative writing and content generation\n\n### ** Performance Matrix (September 2025)**\n\n| Use Case | Top Model | Alternative | RAM Required | Best Profile |\n|----------|-----------|-------------|--------------|--------------|\n| **General Chat** | `llama3.3:70b` | `glm4:latest`  | 64GB / 9GB | Balanced |\n| **Code Development** | `deepseek-coder:33b` | `magicoder:latest`  | 32GB / 7GB | Code |\n| **Reasoning Tasks** | `deepseek-r1` | `glm4:latest`  | 16GB / 9GB | Reasoning |\n| **Creative Writing** | `gemma2:27b` | `gemma3n:latest`  | 32GB / 8GB | Creative |\n| **Resource-Limited** | `granite3.3:2b`  | `gemma3:270m`  | 2GB / 300MB | Balanced |\n| **Vision/Multimodal** | `llava:latest` | `gemma3n:latest`  | 16GB / 8GB | Technical |\n| **Enterprise/128K Context** | `granite3.3:8b`  | `llama3.1:8b` | 8GB | Technical |\n| **Edge/Mobile** | `gemma3:270m`  | `granite3.3:2b`  | 300MB / 2GB | Balanced |\n\n### ** Quick Download Commands (September 2025)**\n```bash\n# Most recommended overall (2025 flagship)\nollama pull llama3.3:70b-instruct\n\n# High-performance alternative (ranks 3rd overall)\nollama pull glm4:latest\n\n# Specialized coding assistant with OSS-Instruct training\nollama pull magicoder:latest\n\n# Premier coding powerhouse\nollama pull deepseek-coder:33b\n\n# Enterprise model with 128K context\nollama pull granite3.3:8b\n\n# Ultra-efficient for mobile/edge (270M parameters)\nollama pull gemma3:270m\n\n# Multimodal for everyday devices\nollama pull gemma3n:latest\n\n# Advanced reasoning powerhouse\nollama pull deepseek-r1\n\n# Vision and image analysis\nollama pull llava:latest\n\n# Best for beginners/limited hardware\nollama pull mistral:7b-instruct\n```\n\n### ** 2025 Community Insights**\n- **Llama 3.3** has become the gold standard for local deployment\n- **DeepSeek models** dominate coding and reasoning benchmarks\n- **Mistral 7B** remains the go-to recommendation for newcomers\n- **Vision models** like LLaVA are gaining massive adoption\n- Over **1,700+ models** now available in Ollama ecosystem\n\n*Note: Model availability and performance may vary. Check [ollama.com/library](https://ollama.com/library) for the latest models.*\n\n---\n\n**Repository**: `universal-ollama-optimizer` | **Developer**: JTGSYSTEMS.COM | **Technology**: Bash, Linux, Ollama, AI\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjtgsystems%2Funiversal-ollama-optimizer","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjtgsystems%2Funiversal-ollama-optimizer","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjtgsystems%2Funiversal-ollama-optimizer/lists"}