Elon Musk reversed his position on Anthropic on July 10, 2026, posting on X that he was “clearly wrong” about the company and that no competitor had released a model as capable as Claude Fable 5 or Mythos 5 — a direct about-face from calling the company’s models “misanthropic and evil” just five months earlier.
Musk’s full post, published on X at approximately 11:00 PM ET on July 10, 2026:
“I was clearly wrong about Anthropic. They are obviously currently the leader in AI. No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon. And I would never cut them off in a way that hurt them badly, even as a competitor.” — Elon Musk, X, July 10, 2026
Anthropic has not publicly responded to the post as of July 20, 2026.
What Musk Said About Anthropic Before This Week
Musk’s July 10 post reversed two years of public attacks on Anthropic. In September 2025, Musk wrote that “winning was never in the set of possible outcomes for Anthropic,” according to TechCrunch’s reporting on July 9, 2026. As recently as February 2026, Musk called Anthropic’s Claude models “misanthropic and evil,” per Yahoo Finance.
The reversal arrived one day after xAI launched Grok 4.5 on July 8, 2026 — a model xAI positioned as a direct coding and agentic competitor to Claude Fable 5. Independent benchmark results already placed Grok 4.5 behind Fable 5 on most measures, according to TechCrunch. Artificial Analysis, the independent AI evaluation firm, placed 3 Anthropic models at the top of its rankings as of July 2026, per The Motley Fool.
What the Benchmark Data Shows
Fable 5 and Mythos 5 launched on June 9, 2026. The U.S. government briefly restricted export of the models under a Trump administration executive order before restoring access on July 1, 2026, as Anthropic confirmed in its official release notes. Fable 5’s SWE-Bench Pro score stands at 80.3%, compared to GPT-5.5 at 58.6% — a 21.7-percentage-point gap on the standard software engineering evaluation, according to Anthropic’s benchmarks.
Musk’s post gave no explanation for his change of view beyond the benchmark results. His post cited Fable and Mythos by name, referencing both the mid-tier and frontier model releases rather than Anthropic’s full product suite.
What the “Cut Them Off” Comment Means
Musk’s pledge not to “cut them off in a way that hurt them badly” refers to potential compute access restrictions. Anthropic runs its models on Amazon Web Services and Google Cloud, not on xAI’s infrastructure. Musk’s actual compute leverage over Anthropic is minimal. When asked directly whether he would weaponize xAI’s resources against Anthropic, Musk wrote “Not my style,” according to Yahoo Finance. The commitment is largely symbolic — but it removes a talking point competitors could use against Anthropic in enterprise sales.
Anthropic co-founder Dario Amodei and co-founder Daniela Amodei both worked at OpenAI, where Musk served as an early board member. Musk’s previous attacks on Anthropic occurred against the backdrop of his own competing AI venture — xAI, which launched Grok models as direct Anthropic rivals. The context makes the July 10 post notable as competitive acknowledgment, not just personal opinion.
- Grok 4.5 Is Here: SpaceXAI’s Opus-Class AI at a Fraction of the Price — the Musk-owned competitor Anthropic is now outperforming
- Fable 5 Is Back: US Lifts Export Controls After Anthropic Adds AI Safety Classifier — the 18-day export ban that preceded Musk’s reversal
- Claude Fable 5 Rewrote Bun in Rust — 1 Million Lines for $165K — the coding benchmark that made Fable 5’s lead concrete
Musk praised Anthropic the same week his own company launched a competing model. The benchmarks made that position untenable. Fable 5 leads the SWE-Bench Pro chart by 21.7 percentage points over GPT-5.5. At that gap, saying otherwise would damage credibility with developers who run their own evaluations. The reversal tells you more about where the benchmark wars stand than it does about Musk — Fable 5 is currently the model every other lab is measured against.
Frequently Asked Questions
What did Elon Musk say about Anthropic on July 10, 2026?
Musk posted on X that he was “clearly wrong about Anthropic” and called the company “obviously currently the leader in AI,” citing Claude Fable 5 and Mythos 5 as the best models any company had released. He also stated he would not cut Anthropic off from compute resources, even as a competitor.
What did Musk say about Anthropic before his reversal?
In September 2025, Musk wrote that “winning was never in the set of possible outcomes for Anthropic.” In February 2026, he called Anthropic’s Claude models “misanthropic and evil.” The July 10, 2026 post reversed both positions within a single statement.
How does Claude Fable 5 compare to other AI models on benchmarks?
Fable 5 scored 80.3% on SWE-Bench Pro, compared to GPT-5.5 at 58.6%, according to Anthropic’s official benchmark disclosures. Artificial Analysis independently placed 3 Anthropic models at the top of its rankings as of July 2026. Grok 4.5, launched by xAI on July 8, 2026, ranked below Fable 5 on most evaluated measures.
Why did Musk say he wouldn’t “cut Anthropic off” from compute?
Musk was responding to questions about whether xAI could restrict Anthropic’s compute access. Anthropic runs Claude models on Amazon Web Services and Google Cloud — not on xAI’s infrastructure. Musk’s leverage over Anthropic’s compute is minimal, making the pledge primarily symbolic in commercial terms.
