Issues

• [Bug]: 0.8.1 — a max_duration run can hang after its last request when the constraint stops queuing (same race as #1475/#1476; please consider a patch release)

  vllm-project/guidellm#1536 · Filter by author danielxeal opened on Nov 2, 2026

• api_routes values with a leading slash produce //v1/... URLs, which the mock server answers with 404

  vllm-project/guidellm#1535 · Filter by author wguo7 opened on Nov 2, 2026

• The image count metric is always empty because image_count is never set

  vllm-project/guidellm#1533 · Filter by author wguo7 opened on Nov 1, 2026

• [Bug]: Without usage in the response, output token counts fall back to the requested target, so early-stopped streams report inflated throughput

  vllm-project/guidellm#1531 · Filter by author Thisish3 opened on Oct 31, 2026

• guidellm run --output kind=console prints every console table twice

  vllm-project/guidellm#1529 · Filter by author wguo7 opened on Oct 31, 2026

• Metrics with no observations are shown as 0.0 in the console and HTML report

  vllm-project/guidellm#1522 · Filter by author wguo7 opened on Oct 30, 2026

• [Bug]: Credentials passed in extras.headers are written in plain text to the reports and the console

  vllm-project/guidellm#1509 · Filter by author wguo7 opened on Oct 28, 2026

• Support native Basal/System One decision benchmarks

  vllm-project/guidellm#1462 · Filter by author Butterski opened on Oct 26, 2026

• [Bug]: Per-modality usage tokens are counted twice in prompt/output token totals

  Bug

  vllm-project/guidellm#1456 · Filter by author saidurpulok opened on Oct 25, 2026

• synthetic_image and synthetic_video send the same media in every benchmark of a run

  vllm-project/guidellm#1454 · Filter by author boqiny opened on Oct 25, 2026

• [Bug]: Trace replay prompts lose one token per hash block with tokenizers that add a BOS token

  vllm-project/guidellm#1424 · Filter by author wguo7 opened on Oct 22, 2026

• [Bug]: Responses API streaming counts function_call_arguments.done as an extra token iteration

  vllm-project/guidellm#1390 · Filter by author wguo7 opened on Oct 20, 2026

• [Bug]: Request-count warmup and cooldown follow completion order, not start order, under concurrency

  vllm-project/guidellm#1379 · Filter by author tomron opened on Oct 20, 2026
  1

• [request] Ability to pass the API token via a file

  Feature

  vllm-project/guidellm#1378 · Filter by author kpouget opened on Oct 20, 2026

• [Bug]: synthetic_image with images_per_request > 1 splits a request's images into separate turns

  vllm-project/guidellm#1337 · Filter by author Jackmeson1 opened on Oct 16, 2026

• [Bug]: /v1/responses: TTFT is taken at the first answer token for vLLM reasoning models

  Bug

  vllm-project/guidellm#1307 · Filter by author wguo7 opened on Oct 14, 2026

• [Bug]: Request-count warmup and cooldown each exclude one request too many

  Bug

  vllm-project/guidellm#1306 · Filter by author wguo7 opened on Oct 14, 2026

• [Bug]: JSON/YAML output with a path not ending in .json/.yaml/.yml fails after the benchmark finishes and no outputs are written

  Bug

  vllm-project/guidellm#1286 · Filter by author JoeyTan21 opened on Oct 10, 2026 · v0.8.2

• guidellm explain

  featureinternalneeds definitionpriority-low
  Feature

  vllm-project/guidellm#1277 · Filter by author dbutenhof opened on Oct 9, 2026 · v1.0.0

• CLI revamp completion

  internalpriority-low
  Feature

  vllm-project/guidellm#1274 · Filter by author dbutenhof opened on Oct 9, 2026 · v1.0.0

• Workload catalog support in GuideLLM

  priority-high
  Feature

  vllm-project/guidellm#1244 · Filter by author ashishkamra opened on Oct 6, 2026 · v1.0.0

• Reported TTFT includes time spent inside GuideLLM

  vllm-project/guidellm#1238 · Filter by author QHarshil opened on Oct 5, 2026 · v1.0.0

• Rearchitect data generation

  featureneeds definition
  Task

  vllm-project/guidellm#1234 · Filter by author dbutenhof opened on Oct 5, 2026 · v1.0.0
  

• Maximum concurrency controls for trace data

  featureinternalpriority-critical
  Feature

  vllm-project/guidellm#1232 · Filter by author dbutenhof opened on Oct 5, 2026 · v0.8.2
  

• Switch synthetic text to a corpus generator

  featurepriority-high
  Feature

  vllm-project/guidellm#1229 · Filter by author dbutenhof opened on Oct 5, 2026 · v1.0.0