# AutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines

arXiv: https://arxiv.org/abs/2609.18123 · HTML: https://mlsanity.com/paper/2609.18123/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 18 of its 23 references that are in the map.
- First posted on arXiv 16 Sep 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)
- Efficient LLM inference and serving / [Efficient reasoning and agent inference](https://mlsanity.com/inference/efficient-reasoning-and-agent-inference/)

## Builds on (most cited first, 18 of 18)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [AlphaEvolve: A coding agent for scientific and algorithmic discovery](https://mlsanity.com/paper/2506.13131/) (2025)
- [EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty](https://mlsanity.com/paper/2401.15077/) (2024)
- [EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees](https://mlsanity.com/paper/2406.16858/) (2024)
- [EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test](https://mlsanity.com/paper/2503.01840/) (2025)
- [KernelBench: Can LLMs Write Efficient GPU Kernels?](https://mlsanity.com/paper/2502.10517/) (2025)
- [Ansor: Generating High-Performance Tensor Programs for Deep Learning](https://mlsanity.com/paper/2006.06762/) (2020)
- [MLPerf Inference Benchmark](https://mlsanity.com/paper/1911.02549/) (2019)
- [SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills](https://mlsanity.com/paper/2308.16369/) (2023)
- [DistillSpec: Improving Speculative Decoding via Knowledge Distillation](https://mlsanity.com/paper/2310.08461/) (2023)
- [Learning to Optimize Tensor Programs](https://mlsanity.com/paper/1805.08166/) (2018)
- [Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization](https://mlsanity.com/paper/2509.14279/) (2025)
- [Astra: A Multi-Agent System for GPU Kernel Performance Optimization](https://mlsanity.com/paper/2509.07506/) (2025)
- [Do Repetitions Matter? Strengthening Reliability in LLM Evaluations](https://mlsanity.com/paper/2509.24086/) (2025)
- [KernelBench-Verified: Do LLM-Generated Kernels Actually Beat PyTorch?](https://mlsanity.com/paper/2607.16241/) (2026)

## Abstract (arXiv)

Large language model agents tune GPU kernels and serving engines through a closed loop of propose, measure, and keep, but the measurements behind this loop are not trustworthy. We characterize four failure modes from a four-day pilot corpus of 619 model calls: strawman baselines manufacture speedups, absolute times do not transfer across machines, saturated tasks nullify comparisons, and infrastructure defects impersonate science. We present AutoTuneBench, a benchmark and measurement protocol that makes trust architectural. The protocol is frozen as code with test-enforced provenance; a database-level validator rejects out-of-protocol results; anti-cheat checks run outside the agent's modification surface; comparisons follow pre-registered readouts; and measurements anchor to externally published results, grounded in paired-seed statistics with a 5% cross-run coefficient-of-variation cap. Honest measurement rewrites the headlines: our best kernel reads 10.6x against a naive baseline but 2.03x against the honest one; one configuration delivers 1.174x on one machine and 1.0049x on another; a pre-registered on/off comparison nulls at a shared wall (2.4840 vs 2.4957,ms); and the KernelBench Level-1 suite admits 51% of tasks with median speedup 1.0001x over PyTorch eager. The protocol, the two-engine corpus (vLLM and SGLang), and its audit trail are released as open artifacts.