This is a plan, not a write-up. I haven’t run Ollama 0.40.16 or Ripple Preview yet. After the 0.40.15 notes and this week’s 0.40.16 release, I want to see whether the new coding model and the Qwen memory cut are worth a daily-driver change.
The plan
- Upgrade Ollama to 0.40.16 in a disposable environment first, not on my daily setup, and confirm the new version starts cleanly.
- Pull Ripple Preview (
ollama run ripple-preview) and write down download size, RAM use at my usual context length, and tokens per second. - Run the same short coding and tool-use prompts through Ripple Preview and a model I already know, and compare answers without reading too much into a small sample.
- On Apple silicon, load Qwen3.8 Flash Next and measure memory against the earlier Qwen3.8 build, looking for that claimed ~27 GiB drop.
- Record versions, quant names and hashes so the numbers can be reproduced, then roll the test environment back.
Questions I want answered
- Does Ripple Preview feel like a real coding model at this size, or just a preview name?
- Is the Qwen memory cut close to 27 GiB in practice, and does quality hold?
- Is any of this stable enough to keep, or is it still “try it and revert”?
Review before you install: read the release notes yourself, back up your config, and try new versions somewhere you can afford to break first.
Isolate it: try new models and tools in a disposable VM or a spare machine first, with a snapshot you can roll back to. Don’t point them at a machine you care about.
God Bless, and Use the Source!
