Search results

[Improvement] output_throughput and output_inference_speed are identical, can remove one of them

Status: Open.

Bug

#202 In sgl-project/genai-bench;

· key4ng opened on Mar 14, 2026

[Bug] Outlier throughput values skew aggregated metrics

Status: Open.

Bug

#201 In sgl-project/genai-bench;

· key4ng opened on Mar 14, 2026

[Proposal] Consider adding p1, p5, p10 metrics for throughput

Status: Open.

#200 In sgl-project/genai-bench;

· key4ng opened on Mar 14, 2026

[Design] Multi-Turn Conversation Support

Status: Open.

#194 In sgl-project/genai-bench;

· OpheliaLjh opened on Mar 5, 2026

[Feature] Output reasoning tokens in the report

Status: Open.

Feature

#162 In sgl-project/genai-bench;

· shenoyvvarun opened on Jan 29, 2026

[Feature] Support benchmarking for multi-modality models

Status: Open.

Feature

#140 In sgl-project/genai-bench;

· shenoyvvarun opened on Dec 15, 2025

[feature] request mode and batch size mode

high priority

Status: Open.

#92 In sgl-project/genai-bench;

· zhyncs opened on Oct 14, 2025

[Proposal] Accept space-separated lists for --num-concurrency and --batch-size (use nargs=-1)

Status: Open.

#35 In sgl-project/genai-bench;

· junliu-mde opened on Jul 7, 2025

[Question] Plan for prompt replay?

Status: Open.

#8 In sgl-project/genai-bench;

· binarycrayon opened on Jun 26, 2025