NVIDIA blog, MLCommons results and supplemental PDF
| Artifact | Date | Role |
|---|---|---|
| NVIDIA blog: Vera Rubin NVL72 MLPerf debut | 16 Sep 2026 | Vendor narrative, entry IDs and ratio claims |
| MLCommons v6.1 results announcement | 16 Sep 2026 | Round rules, new tests, participation record |
| Supplemental PDF (submitter statements) | Embargo lifted 16 Sep 2026 8:00 AM PT | NVIDIA ~300-word statement and partner statements |
| MLCommons visualizer and Datacenter and Edge results pages | Living | Machine-readable Closed Division rows |
The NVIDIA blog footnote places the Vera Rubin claims cited below in MLPerf Inference v6.1, Closed Division.
NVIDIA Vera Rubin NVL72 preview claims (as stated)
From the NVIDIA blog and the NVIDIA block in the supplemental PDF:
| Workload | Software path | Claim vs GB300 NVL72 | Scenarios named |
|---|---|---|---|
| Qwen3-VL | vLLM with NVIDIA Dynamo | up to 3.7x higher throughput | offline, server, interactive |
| DeepSeek-R1 | TensorRT-LLM | up to 2.5x higher token throughput | offline, server, interactive |
NVIDIA cites entry IDs 6.1-0106 and 6.1-0074 for the platform comparison figures.
Hardware and technique claims attached to these submissions in the NVIDIA blog (not separately measured here):
- Enhanced Tensor Cores and Transformer Engine on prefill and decode.
- NVFP4 precision across model weights, attention, and KV cache.
- Disaggregated serving (separate prefill and decode).
- Large-scale expert parallelism for MoE layers.
- NVL72 scale-up via sixth-generation NVLink and NVLink Switch (NVIDIA states 10x higher packet rates and 3x lower latency than off-the-shelf Ethernet in this post).
Partner note: Nebius also submitted Vera Rubin NVL72 preview results (NVIDIA blog; Nebius supplemental names Nebius VR200 NVL72 built on NVIDIA Vera Rubin NVL72).
Companion GB300 NVL72 figures in the same round
This article also cites these figures for the baseline rack:
| Claim | Detail | Entries and notes |
|---|---|---|
| DeepSeek-R1 scale-out | 72 GPUs → 288 GPUs (four GB300 NVL72 racks); 99% offline scaling efficiency vs single-rack baseline | NVIDIA cites 6.1-0073 and 6.1-0074 |
| Qwen3-VL software gain | up to 1.6x vs MLPerf Inference v6.0 on GB300 | lower KV-cache precision, fusion, kernels, disaggregated serving with vLLM and Dynamo |
| WAN 2.2 text-to-video | 0.65 720p videos per second; 5.7 seconds per video on full-rack GB300 NVL72; 9x throughput and 7.5x lower latency vs single-node | NVIDIA blog and supplemental PDF |
| Edge-Agentic | Qwen3.6-27B on Jetson AGX Thor with TensorRT Edge-LLM | new v6.1 edge test |
Post-submission (NVIDIA blog): further unverified gains on GPT-OSS-120B and DLRMv3. MLCommons had not verified them as of the blog.
The SemiAnalysis AgentX preview line in the NVIDIA blog (30x vs GB300) sits outside the Closed Division table. Do not merge it into MLPerf citations.
A record 30 organizations in the v6.1 round
- Results date: 16 September 2026.
- Participation: record 30 organizations (six first-time submitters named by David Kanter).
- New tests: End-to-End RAG and Edge Agentic Inference.
- Speculative decoding supported in interactive for selected benchmarks (including GPT-OSS task).
- Preview and new accelerators named include NVIDIA Rubin and NVIDIA Vera Rubin NVL72.
- Largest system submitted: 512 accelerators (AMD and Crusoe path in supplemental narrative).
-
50% of submitters used the new API-centric harness (foundation for MLPerf Endpoints).
- Per-accelerator server-scenario improvements cited by MLCommons: VLM best result 2.99x vs v6.0 (six months); DeepSeek R1 best result 5.7x vs v5.1 (one year). These are round-level bests. They are separate from Vera Rubin-specific ratios.
Cite entry IDs 6.1-0106 and 6.1-0074
- For Rubin vs GB300 ratios, cite the NVIDIA blog and entry IDs 6.1-0106 and 6.1-0074. Confirm the numeric cells on the MLCommons datacenter results page or visualizer.
- For round rules and new tests, cite the MLCommons announcement.
- For submitter self-description, quote the Supplemental PDF NVIDIA block (and Nebius if partner preview is in scope).
- Mark any power, price, or availability figure as not disclosed unless it appears in these artifacts. The cited passages give no Vera Rubin list price or TDP.
No Vera Rubin list price, TDP or GA date
This article does not include:
- Full per-scenario numeric table reproduced in this article (retrieve live from MLCommons pages; cells can be updated after errata).
- Vera Rubin NVL72 public list price, TDP, and GA date.
- Independent re-implementation of the Dynamo with vLLM or TensorRT-LLM recipes outside submitters.
- Whether post-submission GPT-OSS and DLRMv3 gains will appear in a future verified round.