SpaceX has the best frontier AI model by the end of 2027

Jev's odds

19%chance it comes true

Down 14 points since the first reading on 29 Sep 2026.

How it's read

1. 129 Sep 2026 · First reading. Adds public news: once a week an AI with web search gathers dated developments on the call, Jev keeps the ones that bear on it, and reads them with the evidence on this site. (jev-1.13.0)
2. 230 Sep 2026 · New method. Adds a research brief from Claude Opus 5.5 with web search (weekly, twice weekly in the last 100 days): key numbers and financials, outside odds from prediction markets and analysts, dates coming up, the base rate, and the strongest case each way. Jev reads it with the news and the evidence on this site. (jev-1.13.0)

The odds move when the evidence does, and also when what Jev is given or Jev itself changes, so each change is marked. A model's reading, not a forecast; it can be wrong.

Research brief

From outside this site, by Claude Opus 5.5 with web search on 20 Oct 2026. Jev reads it every day.

Outside odds

• Manifold (play money): Company with the #1 Arena text model on Oct 31, 2026: Google 86%, Anthropic 7%, OpenAI 3%, xAI 2%Manifold, 5 Oct 2026
• Manifold (play money): Top model on the Epoch Capabilities Index at end of 2026: OpenAI 34%, Google 30%, Anthropic 22%, xAI 5%Manifold, 10 Dec 2025
• Manifold (play money): About 31% that a single lab takes a clear, lasting capabilities lead before 2028. Commentary on the market calls xAI 'the smaller challenger.'Tech Insider, 30 Sep 2026
• Elon Musk: Grok 4.7 is roughly on par with Opus 5.0, not 5.1; Grok 4.9 may reach Astra/Fable level; Grok 5 'maybe better than anything.'TeslaNorth, 14 Sep 2026

Key numbers

Grok 4.7 score vs leader on Artificial Analysis Intelligence Index

Grok 4.7 has 46. Claude Opus 5.5 leads with 58; Claude Sonnet 5.5 has 56; Claude Fable 5.1 and GPT-6 Astra have 53. Grok

Codersera · 6 Oct 2026

Best xAI model's place on Arena text leaderboard

#36. The leader, Google's gemini-4-argon-high, had 1525.

Manifold · 2 Oct 2026

Planned Colossus GPU count

About 1.44 million Nvidia GPUs by late December 2026

Tech Insider · 18 Oct 2026

Google GPU lease from SpaceX

About 110,000 GPUs at $920 million per month, October 2026 to June 2029

Wikipedia (Colossus data center) · 5 Jun 2026

SpaceX Q2 2026 revenue / AI segment revenue

$7.8B total revenue (up 92%); $2,561M AI segment revenue

SEC Form 8-K · 30 Jun 2026

SpaceX market value

About $2.2 trillion, with SPCX near $162

TeslaNorth · 9 Oct 2026

Coming up

• 20 Oct 2026SpaceX Q3 2026 results after market close, with a webcast at 3:30 p.m. CT. Watch for AI segment and Grok roadmap news.
• 31 Oct 2026Manifold market on which company holds the #1 Arena text model at the end of October resolves.
• 31 Dec 2026Checkpoints at the end of 2026: the Manifold ECI top-model market resolves, and a Grok 5 release before 2027 is uncertain.
• 30 Jun 2027Musk's SpaceX shares stay locked up until June 2027; there is no fixed date for Grok 4.8, 4.9 or 5.

Base rate. Between August 2024 and October 2026, LMArena published 190 text leaderboards and the lab at #1 changed 7 times, with Google holding it most often. xAI has led only briefly, for example with Grok 3 in early 2025. Grok 5 has slipped from late 2025, to Q1 2026, to Q2 2026, and is still unreleased.

Written by an AI model from what it found on the web; every figure links to its source. It can be wrong or out of date. Not financial advice.

In the news

Public developments that bear on this call, found with the research brief and kept by Jev, which also says which way each points.

• 8 Oct 2026 · AI Weekly / Gizmodo Points against it

  Musk said Grok Bot will now hand tasks to whichever back-end model suits them best, naming Anthropic's Claude Opus 5.5, Midjourney and Suno. Gizmodo read this as SpaceX looking at paths other than leading on frontier models. Source

• 7 Oct 2026 · TVC News / Yahoo Finance Points against it

  Musk said SpaceX will route each Grok Bot task to whichever back-end model is most likely to give the best outcome, Anthropic's Claude Opus 5.5 included, so SpaceX will not rely only on Grok. Source

• 30 Sep 2026 · Essentras LLM Leaderboard Points against it

  Google's model has led LMArena's text leaderboard since September 30, 2026. As of October 15, Gemini 4 Argon (High) was first with 1,524 points, 15 ahead of Meta's Muse Wonder (Max). Source

• 25 Sep 2026 · BigGo Finance Points toward it

  Elon Musk said SpaceX could develop an AI model comparable to Anthropic's Fable or OpenAI's GPT-6 within two to three months and could take the leading position in AI within about six months if its improvement pace holds, while adding up to 660,000 Nvidia GB300 GPUs to reach roughly 1.44 million total GPUs across its Colossus data centers by year-end. Source

• 24 Sep 2026 · Benzinga Points toward it

  Musk said he is 'cautiously optimistic' SpaceX will have a Fable/GPT-6 level model in 2 to 3 months, and that SpaceX will reach 'pole position in about 6 months' if its 'second derivative remains strong'. Source

• 24 Sep 2026 · Forbes Points toward it

  Elon Musk said he was 'cautiously optimistic' SpaceX would have a Fable/GPT-6-level AI model in two to three months and could reach 'pole position' in AI in about six months. Source

• 23 Sep 2026 · Artificial Analysis / DeepLearning.AI The Batch Points against it

  Claude Opus 5.5 took the top spot on the Artificial Analysis Intelligence Index with 58, five points above Claude Fable 5.1 and GPT-6 Astra. Source

• 23 Sep 2026 · Fello AI Points against it

  The Q1 2026 window Elon Musk originally targeted passed, the Q2 (May to June) window xAI later pointed to closed with no Grok 5 release, and the Q3 window is down to its last week. As of September 23, 2026 there is still no confirmed date. Source

• 21 Sep 2026 · Artificial Analysis / The Decoder Points against it

  SpaceXAI released Grok 4.7, which scored 46 on the Artificial Analysis Intelligence Index, putting SpaceXAI among the top 4 labs but behind Claude Fable 5.1 and GPT-6 at 53. Source

• 21 Sep 2026 · BeInCrypto / Artificial Analysis Points against it

  SpaceXAI shipped Grok 4.7, and independent scores place it fourth among frontier AI labs. Artificial Analysis, an independent benchmarking firm, scored the model at 46 on its Intelligence Index, 2 points above Grok 4.6. Claude Fable 5.1, GPT-6 Astra, and Claude Opus 5 all rank higher. Source

• 21 Sep 2026 · Artificial Analysis / The Decoder Points against it

  SpaceXAI released Grok 4.7, and independent tester Artificial Analysis said Grok 4.7 scores 46 on the Artificial Analysis Intelligence Index to bring SpaceXAI into the top 4 AI labs. The Decoder reported it scores just 46 points, landing mid-pack and well behind Claude Fable 5.1 and GPT-6 at 53 each. Source

• 21 Sep 2026 · OrcaRouter Points against it

  SpaceXAI released Grok 4.7, which scored 46 on Artificial Analysis's Intelligence Index (16th of 655 models) — two points above Grok 4.6 but still seven points below the top-scoring Claude Fable 5.1 and GPT-6 Astra models. Source

Timeline

1. 29 Sep 2026 · The call

   SpaceX has the best frontier AI model by the end of 2027

2. 31 Dec 2027 · Settles

   I'll mark it right or wrong once it's settled.

Posts are tied to this call automatically by an AI model (Jev) and can be wrong. Not financial advice.