A fast, cross-platform Go CLI to create symbolic links between Ollama and LM Studio, sharing models bidirectionally without duplicating gigabytes of storage space.
- 🔄 Bidirectional Linking: Link Ollama models to LM Studio OR LM Studio models to Ollama.
- 🧩 Sharded GGUF Support: Automatically groups and links multi-file models (
00001-of-0000N.gguf). - 👁️ Vision & Multimodal Adapters: Pairs and links multimodal projector adapters (
mmproj-*.gguf). - 🌐 Registries & Namespaces: Full support for Hugging Face (
hf.co/...), custom namespaces, and standard models. - 📊 Storage Analytics (
status): Measures active links, detects broken symlinks, and calculates deduplicated disk savings. - 🧹 Dual-Directory Cleanup (
cleanup): Scans both LM Studio and Ollama blobs to find and wipe broken ghost-links. - 🛡️ Safe & Interactive (
delete): Selectively remove links without touching original weights. Never overwrites existing files. - 💻 Cross-Platform: Works natively on macOS, Linux, and Windows (use
--hardlinkswhere symlinks are not permitted).
curl -fsSL https://raw.githubusercontent.com/qaribhaider/ollama-to-lmstudio-symlinks/main/install.sh | bashDownload pre-compiled binaries for macOS, Linux, or Windows from the Releases page:
chmod +x ollama-symlinks-*
sudo mv ollama-symlinks-* /usr/local/bin/ollama-symlinksRun the interactive terminal menu:
ollama-symlinksThe interactive UI allows you to view storage savings, link models forward or reverse, or clean up links with arrow-key navigation.
Inspect active links, broken symlinks, and exact disk space saved across both applications:
ollama-symlinks status
# Add --verbose to list individual files and targets
ollama-symlinks status --verboseScans Ollama manifests and creates symlinks inside LM Studio:
# Interactive selection
ollama-symlinks
# Automated (links all discovered models without prompts)
ollama-symlinks --interactive=false
# Dry run (preview without modifying files)
ollama-symlinks --dry-runScans LM Studio GGUFs and registers them with Ollama:
# Interactive selection with default 'lms-' prefix
ollama-symlinks --reverse
# Automated with custom model prefix
ollama-symlinks --reverse --name-prefix="myorg" --interactive=falseScans both LM Studio and Ollama blobs for orphaned symlinks (e.g. after running ollama rm):
# Preview broken links
ollama-symlinks cleanup --dry-run
# Interactively remove broken links
ollama-symlinks cleanupSafely remove symlinks without touching original model data:
# Remove models linked into LM Studio
ollama-symlinks delete --from lmstudio
# Remove models registered in Ollama
ollama-symlinks delete --from ollama| Flag | Default | Description |
|---|---|---|
--interactive, -i |
true |
Launch interactive selection menus (false for automation). |
--reverse |
false |
Enable Reverse Mode: Link LM Studio models to Ollama. |
--name-prefix |
lms |
Prefix for models imported into Ollama (e.g. lms-llama-3). |
--ollama-dir |
Auto | Path to Ollama models directory (default: ~/.ollama/models). |
--lmstudio-dir |
Auto | Path to LM Studio models directory (default: ~/.cache/lm-studio/models). |
--skip-provider |
ollama |
Folder inside the LM Studio models directory that holds the links. Used by forward mode, status, delete and cleanup, and skipped by reverse mode. Must be a plain folder name. |
--hardlinks |
false |
Use hard links instead of symlinks (resolves Windows permissions / 0-byte issues). Fails with an error if a hard link cannot be made, for example across drives; it never falls back to symlinks. |
--skip-checks |
false |
Skip pre-flight executable validation checks. |
--dry-run |
false |
Preview operations without creating or modifying files. |
--verbose |
false |
Display verbose diagnostic logs. |
--deep-scan |
false |
Scan all available Windows drives for model folders. |
--version |
false |
Display binary version. |
ollama-symlinks status— Shows storage savings, link counts, and health.ollama-symlinks cleanup— Scans LM Studio and Ollama for broken links and removes them.ollama-symlinks delete --from <lmstudio|ollama>— Interactively selects and removes symlinks.
To remove the binary from your system:
curl -fsSL https://raw.githubusercontent.com/qaribhaider/ollama-to-lmstudio-symlinks/main/uninstall.sh | sudo bashsudo rm -f /usr/local/bin/ollama-symlinks(Note: Uninstalling the binary leaves your existing symlinks and original model files intact. Use ollama-symlinks delete prior to removal if you wish to clean up created links first).
- 🪟 Troubleshooting & Windows Guide: Windows Developer Mode, Administrator rights, hard link resolution for 0-byte models, and FAQ.
- 🛠️ Contributing Guide: Building from source, project architecture, and developer workflows.
This project is licensed under the MIT License. See LICENSE for details.