MLPerf Training v6.1 adds first LLM post-training benchmark as Rubin and TPUv7 debut
MLCommons has released results for MLPerf Training v6.1, adding the suite's first large-scale LLM post-training benchmark, the consortium said in a press release. The new Qwen3.5-397B workload uses Group Relative Policy Optimization to post-train a model that learns to solve software engineering tasks, including tool calling. A second new benchmark, DLRMv4-HSTU, replaces the earlier DLRM-dcnv2 recommender test.
Participants submitted 153 systems with 250 test results. 57 of the systems, or 37%, were cloud-based, up from 29% in v6.0, and twelve used 1,000 or more accelerators, the largest 10,240. Systems used NVIDIA, AMD and Google accelerators, including NVIDIA Rubin-generation GPUs and Google's TPUv7 Ironwood for the first time.
24 organisations submitted results, among them Azure, CoreWeave, Dell, HPE, Lambda, Nebius, Oracle and Supermicro, with Crusoe taking part for the first time. "Post-training has added a new level of complexity to the architecture of AI training," said Shriya Rishab, co-chair of the MLPerf Training working group.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Silicon Brief or Daily AI Brief.
Also, consider following us on social media:
More from Data Centers
Nov 16 TensorWave signs six-year deal to provide OpenAI 457MW of AMD-powered AI capacity Nov 16 AMD completes World Labs acquisition, making Fei-Fei Li its chief scientist Nov 16 US to require licences for AI chips at overseas data centres that may serve Chinese firms Nov 16 Bedrock Energy raises $24M for AI-directed geothermal drilling rigs Nov 16 Fed's Barr warns of debt and circular financing behind the AI investment boomSilicon Brief
Weekly coverage of AI hardware developments including chips, GPUs, cloud platforms, and data center technology.
Market report
2025 Generative AI in Professional Services Report
Thomson Reuters
This report by Thomson Reuters explores the integration and impact of generative AI technologies, such as ChatGPT and Microsoft Copilot, within the professional services sector. It highlights the growing adoption of GenAI tools across industries like legal, tax, accounting, and government, and discusses the challenges and opportunities these technologies present. The report also examines professionals' perceptions of GenAI and the need for strategic integration to maximize its value.
Read moreYou may also like
llm-d 1.0 Adds Controls for In House AI Inference
Alibaba launches Qwen4-Max, a closed 2.4-trillion-parameter flagship model
Liquid AI Releases LFM3, Six Open-Weight Models for On-Device AI
OpenAI Reports Major GPT-6 Coding Gains in COBOL and Fortran
Perplexity releases pplx-decider-v1.2 for joint text and image decisions
Daily AI Brief: the AI news that matters, in your inbox.