Skip to content

[llm_bench] Add Qwen3-Omni support (MoE and dense models)#4112

Open
avasenin-14 wants to merge 7 commits into
openvinotoolkit:masterfrom
avasenin-14:llm-bench-qwen3-omni
Open

[llm_bench] Add Qwen3-Omni support (MoE and dense models)#4112
avasenin-14 wants to merge 7 commits into
openvinotoolkit:masterfrom
avasenin-14:llm-bench-qwen3-omni

Conversation

@avasenin-14

@avasenin-14 avasenin-14 commented Jul 8, 2026

Copy link
Copy Markdown
Contributor

Description

CVS-188611

Depends on pull requests huggingface/optimum-intel#1700 and huggingface/optimum-intel#1640.

Add UseCaseVLM with OVModelForMultimodalLM for Qwen3-Omni. Add tests for the new model.

Limitations:

  • PyTorch path does not exist for VLMs currently. I can add it also if there is a need.
  • Text-to-audio IR export is not ready for the dense model yet, so I'll add it another PR.

Checklist:

  • This PR follows GenAI Contributing guidelines.
  • Tests have been updated or added to cover the new code.
  • This PR fully addresses the ticket.
  • I have made corresponding changes to the documentation.

@avasenin-14 avasenin-14 requested a review from Copilot July 8, 2026 10:31
@avasenin-14 avasenin-14 self-assigned this Jul 8, 2026
@github-actions github-actions Bot added the category: llm_bench Label for tool/llm_bench folder label Jul 8, 2026

Copilot AI 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.

Pull request overview

Adds llm_bench model-type routing for Qwen3-Omni VLM by selecting OVModelForMultimodalLM when the detected model type is qwen3-omni, enabling Optimum-based VLM benchmarking for this model family.

Changes:

  • Import OVModelForMultimodalLM from optimum.intel.openvino.
  • Extend USE_CASES["visual_text_gen"] with a dedicated UseCaseVLM(["qwen3-omni"], ov_cls=OVModelForMultimodalLM) entry.

Comment thread tools/llm_bench/llm_bench_utils/config_class.py Outdated
Copilot AI review requested due to automatic review settings July 8, 2026 11:52
@github-actions github-actions Bot added the category: GGUF GGUF file reader label Jul 8, 2026

Copilot AI 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.

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

Comment thread tests/python_tests/samples/test_tools_llm_benchmark.py Outdated
Copilot AI review requested due to automatic review settings July 8, 2026 12:02

Copilot AI 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.

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

Comment thread tests/python_tests/samples/test_tools_llm_benchmark.py
Copilot AI review requested due to automatic review settings July 8, 2026 12:23

Copilot AI 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.

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

Comment thread tests/python_tests/samples/conftest.py
@avasenin-14 avasenin-14 marked this pull request as ready for review July 8, 2026 12:33
@avasenin-14 avasenin-14 requested a review from akashchi as a code owner July 8, 2026 14:59
Copilot AI review requested due to automatic review settings July 8, 2026 14:59
@github-actions github-actions Bot added the category: GHA CI based on Github actions label Jul 8, 2026

Copilot AI 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.

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

@avasenin-14

Copy link
Copy Markdown
Contributor Author

Exports of tiny random model have failed in test as optimum-intel package in CI is built from a commit that doesn't have the support for qwen3-omni models yet.
It works locally, so as soon as the pull requests in optimum-intel will be merged, I'll rerun the CI to check it.

Copilot AI review requested due to automatic review settings July 8, 2026 20:03

Copilot AI 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.

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

Copilot AI review requested due to automatic review settings July 9, 2026 10:10

Copilot AI 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.

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

Comment on lines +254 to +260
try:
from optimum.intel.openvino import OVModelForMultimodalLM
except ImportError:
pass
else:
USE_CASES["visual_text_gen"].append(UseCaseVLM(["qwen3-omni"], ov_cls=OVModelForMultimodalLM))

Comment on lines +727 to +732
"sample_args",
[
["-d", "cpu", "-n", "1", "--num_steps", "4", "--task", "visual_text_gen", "--optimum"],
["-d", "cpu", "-n", "1", "--num_steps", "4", "--task", "visual_text_gen", "--genai"],
],
)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

category: GGUF GGUF file reader category: GHA CI based on Github actions category: llm_bench Label for tool/llm_bench folder

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants