The Circuitry
THE CIRCUITRYYour one-stop source for all tech news
  • Today
  • Feed
  • Events
  • Saved
HOMETODAYFEEDEVENTS
BOOKMARKS
RSS
© 2026 The Circuitry
About UsSourcesContactCorrectionsPrivacy
  • Today
  • Feed
  • Events
  • Saved
  • AAPL
  • MSFT
  • GOOGL
  • META
  • NVDA
  • TSLA
  • AMZN
  • INTC
  • SONY
  • TTWO
  • COIN
  • BTC
  • ETH
  • SOL
  • AAPL
  • MSFT
  • GOOGL
  • META
  • NVDA
  • TSLA
  • AMZN
  • INTC
  • SONY
  • TTWO
  • COIN
  • BTC
  • ETH
  • SOL

Feed

Ad
From The CircuitryVerified tech news, cross-checked.How we verify →
  • NVIDIA's third-quarter revenue rises 103% to $115.8 billionTech
  • NVIDIA announces a Vera CPU rack for science workTech
  • Kinsus plans more than 3.5x substrate capacity for NVIDIATech
  • Jensen Huang says he would return as Nvidia CEO if directors decideTech
  • Isomorphic Labs raises $5 billion in a Series CTech
  • Google links COLD RELIC's 2025 US breach to a second intrusionTech
  • Microsoft's Ignite keynote adds Copilot Workspaces and Cobalt 200Tech
  • Google launches a Robotics plugin for Gemini 4 HydrogenTech
  • Apple says an internal update caused today's service outagesTech
  • Google Wallet is now available on iOS in the U.S.Tech
Scroll for more
Verification
VERIFIEDConfidence: HIGH
Source identified
Claims cross-referenced
No discrepancies found
Sourcing
1source
NVIDIA · track record
10Stories
100%Verified
930d
All sources →
Markets
NVDA···

Live quote · not investment advice

Ad
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →
Home/Tech/NVIDIA Vera Rubin leads the new MLPerf training round
VERIFIEDBy Xavier Rivera·Monday, November 16, 2026 at 12:11 PM AST·2 min read

NVIDIA Vera Rubin leads the new MLPerf training round

NVIDIA has posted Vera Rubin scores for MLPerf Training v6.1, and the racks led every test released today. Vera Rubin NVL72 did up to twice the work per chip of GB300 and more than 10 times the agent output per watt.

Source:NVIDIA
Post
NVIDIA Vera Rubin leads the new MLPerf training round
Ad
From The CircuitryVerified tech news, cross-checked.How we verify →
Ad
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →
TL;DRAI · 60 sec read

NVIDIA has posted Vera Rubin scores for MLPerf Training v6.1, and the racks led every test. Vera Rubin NVL72 did up to twice the work per chip of GB300. A larger run kept about 97 percent efficiency and cut a 16-minute job to about four. A first Endpoints filing shows more than 10 times the agent output per watt.

Morning Brief

Get the daily Nvidia brief

Every morning, in your inbox. What's in The Brief →

NVIDIA has posted Vera Rubin scores for MLPerf Training v6.1, and the racks led every test in the round released today.

The new rack beat the prior one on training speed. Vera Rubin NVL72 turned in up to twice the work per graphics chip of a tuned GB300 NVL72. NVIDIA was the only entrant with a filing on each test. The set covers language models, image generation, recommenders, and the reinforcement-learning stage that follows the main run.
NVIDIA was the only entrant with a filing on each test.
Scaling held up as the cluster grew. On DeepSeek-V3, Vera Rubin NVL72 went from 256 chips to 1,024 and kept about 97 percent of the expected speedup. Multiplying the chip count by four cut that job from 16 minutes to about four.

NVIDIA also filed the test on 10,240 GB300 chips, the biggest cluster in the round. Microsoft Azure matched that count with GB200 systems.
From The CircuitryThe Feed — live briefs across tech, all day.See what’s happening →
Agent benchmarks are a separate first for these racks. This round is NVIDIA's first filing in MLPerf Endpoints. Those tests time coding and office agents and check that the agent still completes the assignment. The filing shows more than a tenfold gain over the prior generation in agent work finished for each watt.
Those tests time coding and office agents and check that the agent still completes the assignment.
A provisional Kimi K3 score puts Vera Rubin NVL72 at more than 10 times the output per watt of GB300 NVL72 when the reply speed is held even.

The clock starts at a finished model, not a single step. MLPerf Training measures how long a system needs to hit a target accuracy. Shorter runs mean coding helpers, lab models, and fraud checks can ship sooner. Chips inside a rack talk over NVLink, and racks talk over Quantum-X800 InfiniBand and Spectrum-X Ethernet.
Ad
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →
Partners and outside suites line up with the same result. Twenty groups filed training scores, including cloud operators and server builders. Vera Rubin also leads AgentX and AgentPerf, two agent suites run outside this round. Later software tuning is expected to raise the scores on racks customers already buy.
Why this mattersAI · ~100 words

Tap a lens to see what this story means for you.

Morning Brief

Liked this? The Brief brings you the whole day in tech, verified, every morning.

Two minutes, free forever. What's in The Brief →

Reader-supported
DonateBuy me a coffee →Follow@thecircuitry_ →Follow@thecircuitry.to →
HELP US IMPROVE
Grand Theft Auto VILive hubUS store unlocks Thu Nov 19, 1:00 AM ASTFirst unlock: New Zealand · Wed 7:00 AM ASTLaunch hub →Image: Rockstar Games
From The Circuitry

See what’s happening right now

The Feed runs all day — short, verified briefs the moment they break.

Open the Feed →
From The Circuitry

Follow @thecircuitry_

Every story we publish, as it happens. No noise between.

Follow on X ↗On Bluesky ↗

Reader-supported

The Circuitry is a passion project I've always wanted to build, and I love the work behind it.

Running it costs real money. APIs, hosting, time. To keep improving the site and growing this into something useful for everyone, those costs have to be covered.

Any contribution is appreciated. If not, no pressure. Thanks for reading.

Buy me a coffee
nvidiaaichips
Grand Theft Auto VILive hubUS store unlocks Thu Nov 19, 1:00 AM ASTFirst unlock: New Zealand · Wed 7:00 AM ASTLaunch hub →Image: Rockstar Games
More inTech
  • Google links COLD RELIC's 2025 US breach to a second intrusion

    Tech · 2h
  • Microsoft's Ignite keynote adds Copilot Workspaces and Cobalt 200

    Tech · 2h
  • Google launches a Robotics plugin for Gemini 4 Hydrogen

    Tech · 3h
SupportThe Work

The Circuitry is reader-supported. If you find the daily brief useful, you can buy me a coffee to keep it going.

Buy a coffee →
Ad
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →
From The CircuitryWhy The Circuitry

Verified tech news, cross-checked.

Every story is checked against independent sources before it posts — no rumors dressed up as fact.

How we verify →

MORE IN THIS BEAT

All Tech →
  • Tech·Nov 17

    NVIDIA's third-quarter revenue rises 103% to $115.8 billion

    NVIDIA has reported third-quarter revenue of $115.8 billion, up 103% from a year ago, with data-center revenue of $107.6 billion. Fourth-quarter revenue is expected at $130.0 billion, plus or minus 3%, with no data-center compute revenue from China assumed.

  • Tech·Nov 16

    NVIDIA announces a Vera CPU rack for science work

    NVIDIA has announced a rack of Vera CPUs for scientific computing, built on its first fully custom data-center processor. Dell, HPE, Lenovo, and Supermicro are expected to offer the racks in early 2027.

  • Tech·Nov 17

    Kinsus plans more than 3.5x substrate capacity for NVIDIA

    NVIDIA has expanded supply agreements with Kinsus, Samsung Electro-Mechanics, Amkor, SPIL, Foxconn and TSMC to add manufacturing capacity for its AI platforms. Kinsus will raise dedicated substrate capacity by more than 3.5 times, with production expected to start in the second half of 2027.

  • Tech·Nov 17

    Jensen Huang says he would return as Nvidia CEO if directors decide

    Jensen Huang has said he would return as Nvidia CEO if the board decides a comeback serves shareholders, employees, the company, and the country. He writes that he regrets problems he created for the board and will cooperate with its independent investigation.

  • Tech·Nov 17

    Isomorphic Labs raises $5 billion in a Series C

    Isomorphic Labs has raised $5 billion in a Series C co-led by Greenoaks and Sequoia Capital, with further investment from Alphabet. The company will scale IsoDDE, its AI drug design engine, and move treatment programs toward clinical trials.