Start free trial

TRUE, BUTCherry-Picked Slice

The sample size was ten questions.

“A viral claim on X said the mystery stealth model 'Ox Alpha' beats Claude Fable 5 and GPT-5.6 Sol on the DeepSWE coding benchmark, based on an 8/10-task sample (effectively >80%).”

Ben Davis, @davis7 on X, amplified by AGTP (@AGTPinsights) and Coin Bureau (@coinbureau) to a much larger audience without the sample-size caveat, 21 August 2026

7 receipts2 support2 refute3 context

Issue 1728 August 20263 min

the tester had run 10 of the benchmark's 113 tasks. He ran the rest the next day and landed at 63%. An independent lab ran all 113 and got 58.4%, calling the 80% figure completely incorrect.

The twist

the model's maker, Zhipu AI, has since confirmed its identity as GLM-5.3-Flash, and on a more rigorous multi-trial benchmark the base GLM-5.3 model (not confirmed as the same Flash variant) loses to GPT-5.6 Sol on a single attempt but ties or leads it once retries are allowed.

80%Ben Davis's original viral small-sample score for Ox Alpha on DeepSWE
63%Ben Davis's own full 113-task re-run score
58.4%independent StealthModelWatch full 113-task run
72.7%Together AI's rigorous 4-trial pass@1 for GPT-5.6 Sol on DeepSWE

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 hello@bskiller.com. 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 Cherry-Picked Slice: the flattering subset, presented as the whole. You'll see it again. Learn to spot it →

Say this in tomorrow's meeting“Ten questions is a party trick, not a benchmark.”

Receipts

  1. Supports x.com: gpt-5.6-sol: 52% fable: 65% whatever the hell this is: 80%
  2. Refutes x.com: Actual DeepSWE run on the ox alpha mystery model is done. Ended at ~63% NOT the 80% my first subset test got
  3. Refutes stealthmodelwatch.online: The rumored ~80% pass rate is completely incorrect
  4. Context together.ai: Sol edges ahead: 72.7% pass@1 to GLM-5.3's 69.0%
  5. Supports x.com: BREAKING: A mysterious new AI model called "Ox Alpha" is reportedly BEATING Claude Fable 5 and GPT-5.6 Sol at coding, and nobody knows who built it
  6. Context vktr.com: On August 26, Z.ai confirmed authorship of the model, vindicating the fingerprinting. The model's official name is GLM-5.3-Flash
  7. Context together.ai: It ties Sol at pass@2 (81.1 vs. 81.0) and leads pass@4 (87.6% vs. 85.8%)

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 28 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.