Skip to content

feat(inference): add qwen3.6:35b to Ollama starter models on large-memory hosts#3422

Merged
ericksoa merged 3 commits into
mainfrom
fix/ollama-model-list
May 13, 2026
Merged

feat(inference): add qwen3.6:35b to Ollama starter models on large-memory hosts#3422
ericksoa merged 3 commits into
mainfrom
fix/ollama-model-list

Conversation

@zyang-dev

@zyang-dev zyang-dev commented May 12, 2026 •
edited by coderabbitai Bot
Loading

Copy link
Copy Markdown
Contributor

Summary

Ollama starter-model list now includes qwen3.6:35b alongside qwen2.5:7b and nemotron-3-nano:30b.

Related Issue

Fixes #3250

Changes

  • src/lib/inference/local.ts:
    • New QWEN3_6_OLLAMA_MODEL = "qwen3.6:35b" constant.
    • getBootstrapOllamaModelOptions appends it to the options when gpu.totalMemoryMB >= LARGE_OLLAMA_MIN_MEMORY_MB, after the existing nemotron-3-nano:30b entry.
  • src/lib/inference/ollama/model-size.ts:
    • Added qwen3.6:35b to the FALLBACK_SIZE_BYTES table at 24 GB so the memory probe has an estimate without hitting the manifest.
  • src/lib/inference/local.test.ts:
    • Updated the bootstrap-options test to expect three entries on large-memory hosts.
  • docs/inference/use-local-inference.md:
    • Added one sentence noting that qwen3.6:35b appears in the starter list on hosts with ≥32 GiB of detected GPU memory.

Type of Change

  • Code change (feature, bug fix, or refactor)
  • Code change with doc updates
  • Doc only (prose changes, no code sample modifications)
  • Doc only (includes code sample changes)

Verification

  • npx prek run --all-files passes
  • npm test passes
  • Tests added or updated for new or changed behavior
  • No secrets, API keys, or credentials committed
  • Docs updated for user-facing behavior changes
  • make docs builds without warnings (doc changes only)
  • Doc pages follow the style guide (doc changes only)
  • New doc pages include SPDX header and frontmatter (new pages only)

Signed-off-by: zyang-dev 267119621+zyang-dev@users.noreply.github.com

Summary by CodeRabbit

  • New Features
    • Added Qwen 3.6 (35B) as a starter option for local inference on systems with ≥32 GiB detected GPU memory; becomes the default when no models are discovered on large-GPU hosts.
  • Documentation
    • Noted the Ollama onboarding flow update reflecting the new starter model.
  • Tests
    • Updated test expectations to include the Qwen 3.6 starter option.

Review Change Stack

@zyang-dev
feat(inference): add qwen3.6:35b to Ollama starter models on large-me…
5e220f7 
…mory hosts

Signed-off-by: zyang-dev <267119621+zyang-dev@users.noreply.github.com>
@zyang-dev zyang-dev self-assigned this May 12, 2026
@coderabbitai

coderabbitai Bot commented May 12, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 6f361d13-8e98-49b3-b0ee-867c5b1cb2dc

📥 Commits

Reviewing files that changed from the base of the PR and between 5e220f7 and 947f907.

📒 Files selected for processing (3)
  • docs/inference/use-local-inference.md
  • src/lib/inference/local.test.ts
  • src/lib/inference/local.ts
✅ Files skipped from review due to trivial changes (1)
  • docs/inference/use-local-inference.md
📝 Walkthrough

Walkthrough

This PR adds Qwen3.6 35B as a bootstrap Ollama model option for hosts with ≥32 GiB GPU memory. Changes include the model constant export, fallback size metadata, bootstrap selection and fallback logic, test updates, and user documentation.

Changes

Qwen3.6 Model Addition

Layer / File(s) Summary
Model constant and size metadata
src/lib/inference/local.ts, src/lib/inference/ollama/model-size.ts		 New exported constant QWEN3_6_OLLAMA_MODEL = "qwen3.6:35b" and corresponding fallback byte-size entry for registry probing fallback.
Bootstrap integration and fallback logic + tests
src/lib/inference/local.ts, src/lib/inference/local.test.ts		 getBootstrapOllamaModelOptions pushes QWEN3_6_OLLAMA_MODEL; getDefaultOllamaModel fallback now returns QWEN3_6_OLLAMA_MODEL for large-memory GPUs and SMALL_OLLAMA_MODEL otherwise. Tests import the new constant and update expectations for the LARGE_OLLAMA_MIN_MEMORY_MB case.
User-facing documentation
docs/inference/use-local-inference.md		 Onboarding docs updated to state that with ≥32 GiB detected GPU memory, the starter model list includes qwen3.6:35b alongside smaller defaults.

Sequence Diagram

sequenceDiagram
  participant OnboardWizard
  participant GPUDetector
  participant LocalInference
  OnboardWizard->>GPUDetector: query totalMemoryMB
  GPUDetector-->>OnboardWizard: totalMemoryMB value
  OnboardWizard->>LocalInference: request bootstrap models
  LocalInference->>LocalInference: getBootstrapOllamaModelOptions(gpu)
  LocalInference-->>OnboardWizard: model list (includes qwen3.6:35b when memory >= threshold)
Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Poem

🐰 A rabbit scurries to the gate,
Qwen3.6 hops in—no need to wait,
Thirty-five billion, now in view,
For GPU hosts with memory true. ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name Status Explanation Resolution
Docstring Coverage ⚠️ Warning Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Write docstrings for the functions missing them to satisfy the coverage threshold.
✅ Passed checks (4 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The PR title clearly and accurately summarizes the main change: adding qwen3.6:35b to Ollama starter models specifically for large-memory hosts.
Linked Issues check ✅ Passed All coding requirements from issue #3250 are met: qwen3.6:35b is added to the starter model list, memory estimation is provided, tests are updated, and documentation is updated.
Out of Scope Changes check ✅ Passed All changes are directly scoped to issue #3250: adding qwen3.6:35b to the Ollama starter models on large-memory hosts. No unrelated changes detected.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
📝 Generate docstrings
  • Create stacked PR
  • Commit on current branch
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch fix/ollama-model-list

Comment @coderabbitai help to get the list of available commands and usage tips.

@zyang-dev zyang-dev added v0.0.40 and removed v0.0.40 labels May 12, 2026
ericksoa
ericksoa approved these changes May 13, 2026
View reviewed changes

@ericksoa ericksoa left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved after review and focused validation. Nightly E2E dispatched separately for post-merge signal.

@ericksoa ericksoa merged commit 800d924 into main May 13, 2026
71 of 72 checks passed
@ericksoa ericksoa deleted the fix/ollama-model-list branch May 13, 2026 01:26
@coderabbitai coderabbitai Bot mentioned this pull request May 14, 2026
Merged
12 tasks
@wscurran wscurran mentioned this pull request May 19, 2026
Closed
@wscurran wscurran added the feature PR adds or expands user-visible functionality label Jun 8, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ericksoa ericksoa ericksoa approved these changes

Assignees

@zyang-dev zyang-dev

Labels

feature PR adds or expands user-visible functionality

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

[DGX Spark][Onboard] POR default model Qwen3.6 35B missing from Ollama starter model list

3 participants

@zyang-dev @ericksoa @wscurran