Calliope AI Lab v1.4.0 is out now, and it is the biggest desktop release since launch. A month of focused engineering went into this one: faster notebook generation, a completely overhauled natural language SQL engine, better Ollama support for fully local workflows, and a pile of quality-of-life improvements that add up to a meaningfully better day-to-day experience.

Download Calliope AI Lab v1.4.0

What is Calliope AI Lab?

Calliope AI Lab is a free, locally-running AI data science environment built on JupyterLab. Bring your own API keys for Claude, OpenAI, Gemini, Mistral, or run fully private with Ollama. No account required to get started. It runs on macOS, Windows, and Linux.

What is new in v1.4.0

Notebook Generation is Now 2x Faster

One-shot notebook generation from natural language descriptions has been completely re-architected in this release. The new generation pipeline streams cells as they are produced rather than waiting for the full notebook to complete. For most workflows, you will see your first runnable cell in under 3 seconds. Full notebooks are hitting 2x generation speed across the board.

Overhauled Natural Language to SQL

The natural language SQL engine has been rebuilt from the ground up. It now understands schema context far more reliably, handles multi-table joins correctly on the first attempt, and surfaces a plain-English explanation of every query it generates before running it. For anyone doing exploratory data work across large schemas, this is a substantial upgrade.

New in this release: the SQL assistant will flag queries that touch large row counts before execution and suggest optimized alternatives. No more accidental full-table scans in production databases.

Ollama Integration Overhaul

Running fully offline just got a lot smoother. Lab now auto-discovers Ollama models on startup, with no manual configuration required. Model switching in the AI assistant panel is instant. The new Ollama health indicator shows model status and memory usage right in the sidebar so you always know what is running.

Dataset Registration for Chat Studio

AI Lab notebooks can now register datasets directly to Chat Studio with a single command. Run your analysis, register the output, and your team can start asking natural language questions against it in Chat Studio immediately. This tightens the loop between exploration and sharing considerably.

UI and Workspace Improvements

The cell toolbar has been redesigned to reduce visual noise. AI suggestion overlays no longer obscure adjacent cells. The data source connection panel now shows connection health status inline. Kernel restart and interrupt are more prominent and reliable.

Performance

Cold launch time is down 30% on macOS and Windows. Memory usage during large notebook execution is meaningfully lower thanks to a smarter cell output buffer. Users with older MacBook Pros will notice the difference.

Bug Fixes

• Fixed a race condition in multi-model switching that could cause the assistant to respond with the wrong model
• Fixed an issue where SQLite database previews would fail on databases with no tables
• Fixed cell output truncation for large DataFrames in certain display configurations
• Resolved a startup crash on Windows affecting users with non-ASCII characters in their system username
• Fixed the “Restart and Run All” shortcut incorrectly triggering on some Linux keyboard layouts

Download

Download Calliope AI Lab v1.4.0

Available for macOS (Apple Silicon and Intel), Windows, and Linux. Free for personal and commercial use. Bring your own API keys.

What is Next

The v1.5.0 milestone is focused on collaborative notebook sharing, a new pipeline execution view, and deeper Zentinelle integration for teams that need governance around data access. More soon.