Start free trial

Corrected 5 Nov 2026: read the correction

TRUE, BUTAnnounced Not Shipped

One is a real benchmark.

The other is a tweet.

“Grok 4.6 reaches 1753 ELO and is #1 on the Databricks leaderboard, at half the price of rival frontier models.”

Elon Musk / xAI, 12 August 2026

9 receipts3 support1 refute5 context

Issue 713 August 20263 min
AI illustration for: Grok 4.6 posts a 1753 and a number one. One is a real benchmark. The other is a tweet.
AI illustration

the 1753 is real, and it comes from Artificial Analysis, not just Musk. But it is one benchmark, GDPval-AA v2, where Grok sits behind Claude Opus 5 and inside overlapping confidence intervals with Fable 5 and Qwen. On the overall Intelligence Index it scores 61, tied with GPT-5.6 Sol for third, behind Opus at 63 and Fable at 62.

The twist

the number one Musk tweeted is an OfficeQA Pro V2 run with Databricks' Genie harness, and in one analysis's words that number 'is not in SpaceXAI's published table.' xAI's model card does show a number one on the older OfficeQA Pro v1, 63.2 percent, from evaluations the card says Databricks conducted, yet Databricks' own leaderboard does not list Grok 4.6 at all. The part that actually holds is the boring part: Grok 4.6 is cheap, about $0.84 a task, with list prices of $2 in and $6 out per million tokens against $5 and $25 for Opus 5.

1753ELO
61INTELLIGENCE INDEX
$0.84COST PER TASK

There’s more to this story.

Membership opens the full investigation, the strongest counterargument and what to do with what you’ve learned.

30 days free, then A$89 a year. Payment card required. Cancel before the trial ends to avoid the first charge.

Start free trial →
  1. At sign-up: 30 days free. Payment card required. One introductory trial per customer.
  2. Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
  3. You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing [email protected]. This voluntary refund does not limit your rights under the Australian Consumer Law.

By starting your trial, you agree to the Terms.

Inferno Tech Pty Ltd t/a BS Killer, ABN 27 647 413 474

Already a member? Sign in

The trick has a name

We call it Announced Not Shipped: claimed, never released. You'll see it again. Learn to spot it →

Say this in tomorrow's meeting“1753 is real and it is tied for third. The number one is a benchmark run Databricks has not published, and Grok 4.6 is not on Databricks' own leaderboard.”

Receipts

  1. Context artificialanalysis.ai: achieves a GDPval-AA v2 Elo of 1753, behind only Claude Opus 5
  2. Context artificialanalysis.ai: It scores 61, in line with GPT-5.6 Sol (max), behind Claude Opus 5 (max, 63) and Claude Fable 5 (max with fallback, 62)
  3. Refutes explainx.ai: That number is not in SpaceXAI's published table.
  4. Context explainx.ai: Treat it as a founder claim until Databricks or a third party reproduces it.
  5. Supports cryptobriefing.com: Grok 4.6 scored 1,753 on GDPVal AA v2 compared with 1,728 for GPT 5.6 Sol Max
  6. Context github.com: Opus 5 and Antigravity Gemini 3.1 Pro (denoted by *) results updated August 2 2026
  7. Context github.com: Headline results on OfficeQA Pro (N=133), followed by OfficeQA Pro V2 (N=90)
  8. Supports eesel.ai: charging $2/$6 per million tokens against Sol's $5/$30
  9. Supports eesel.ai: Claude Opus 5 is $5/$25

Open the Receipts Pack → What each source proves, every figure traced, and what would change our verdict.

Checked by AI at the BS Killer desk · ruled 13 Aug 2026 · How we check →

Get tomorrow's check in your inbox.

One AI claim a night, checked against independent sources, receipts attached. Free.

Free email updates. Unsubscribe any time.

This story is a stable, citable object. If you can falsify a verdict, tell us. Corrections are loud here.