# vLLM — User Demand Report

**Week:** 2026-W15
**Generated:** 2026-04-06
**Issues analyzed:** 35 (35 included)
**Need clusters:** 1

## Top 10 User Needs

| Rank | Need | Issues | Score | Category | Examples |
| --- | --- | --- | --- | --- | --- |
| 1 | MoE Performance, Quantization, and Backend Stability Fixes | 35 | 4.5 | Performance | [#39060](https://github.com/vllm-project/vllm/issues/39060), [#39030](https://github.com/vllm-project/vllm/issues/39030), [#39025](https://github.com/vllm-project/vllm/issues/39025) |

## Rising Needs

| Need | Rising Score | This Week | Category |
| --- | --- | --- | --- |
| MoE Performance, Quantization, and Backend Stability Fixes | 36.0x | 35 | Performance |

## Category Breakdown

- **Performance**: 1 clusters

## All Need Clusters

### 1. MoE Performance, Quantization, and Backend Stability Fixes

Users are reporting critical issues with Mixture of Experts (MoE) model performance including significant decode throughput regressions, quantization-related accuracy problems with new models like Gemma 4 and Qwen3, and CUDA/ROCm backend stability issues causing crashes and hangs. These fixes are essential for running large-scale MoE deployments reliably and efficiently.

- **Volume:** 35 issues (31 open, 4 closed)
- **Demand Score:** 4.5
- **Avg Reactions:** 0.1 | **Avg Comments:** 1.3
- **Example issues:** [#39060](https://github.com/vllm-project/vllm/issues/39060), [#39030](https://github.com/vllm-project/vllm/issues/39030), [#39025](https://github.com/vllm-project/vllm/issues/39025), [#39010](https://github.com/vllm-project/vllm/issues/39010), [#39004](https://github.com/vllm-project/vllm/issues/39004)

---

*This report analyzes public GitHub issues only. It represents a signal from public issue discussions, not the full user base.*

*Generated by [ReadYourUsers](https://github.com/study8677/ReadYourUsers)*