DOOMBENCH.COM
SpaceXAI 🇺🇸 · Grok

Grok 4.6

Grok 4.6 is a released general-purpose model optimized for sustained agentic knowledge work, coding, application creation, and other long-running tasks across several hosted platforms.

DOOM SCORE84.8out of 100model risk profile, not the overall index
0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

Send detailed or private feedback →
CURRENT ASSESSMENT · REVISION 2

Why this model scores 84.8

Primary benchmark evidence shows broad gains over Grok 4.5 in coding and agentic work, while API and multi-platform access create high deployment exposure. The release claims expanded predeployment and postdeployment safeguards but provides no detailed system card; official enterprise terms establish access as a proprietary hosted service.

Capability91
Autonomy90
Deployment91
Misuse potential80
Control difficulty69
WATCH AND EXPLORE

Reviewed videos from Artificial Intelligence Videos. These links add context and never change a DoomBench score.

Thumbnail for Grok 4.6 Makes xAI a Serious AI Rival17:15
AI Videos

Grok 4.6 Makes xAI a Serious AI Rival

Matthew Berman tests Grok 4.6 on coding, knowledge work, speed and cost, then explains how Cursor data and xAI compute changed the model race.

The video directly tests Grok 4.6 across coding, knowledge work, speed, and cost.
Thumbnail for Grok 4.6 Gains Intelligence but Loses Its Edge25:31
AI Videos

Grok 4.6 Gains Intelligence but Loses Its Edge

Theo Browne tests Grok 4.6 and finds stronger coding-agent reasoning, but worse token efficiency, slower output and higher real-world task costs.

The video directly tests Grok 4.6 reasoning, token efficiency, output speed, and task cost.
Thumbnail for What Grok 4.6 Benchmarks Miss7:14
AI Videos

What Grok 4.6 Benchmarks Miss

The AI Grid compares Grok 4.6 benchmarks, pricing and early user reports, finding strong value but conflicting evidence about real-world performance.

The video directly evaluates Grok 4.6 benchmark claims against pricing and early practical reports.
Thumbnail for Grok 4.6 Is INSANE – Is THIS a Frontier Model?34:55
AI Videos

Grok 4.6 Is INSANE – Is THIS a Frontier Model?

Bijan Bowen tests Grok 4.6 across coding, games, 3D printing and web design to see where it approaches frontier-model performance.

The video directly tests the exact Grok 4.6 model across several practical frontier-capability tasks.
Thumbnail for Grok 4.6 is Claude Fable 5, but dirt cheap13:19
AI Videos

Grok 4.6 is Claude Fable 5, but dirt cheap

Alex Finn tests Grok 4.6 against other frontier models and finds strong coding value, while its general-purpose agent tools remain less complete.

The review directly compares the exact Grok 4.6 model's coding capability, cost, and agent-tool limitations.
Thumbnail for How Grok 4.6 Follows a Design System38:37
AI Videos

How Grok 4.6 Follows a Design System

Ray Fernando tests whether Grok 4.6 can turn a written design system into a responsive interface through Cursor cloud agents and structured direction.

The demonstration materially illustrates the exact model's instruction following, tool use, and sustained coding work.
MODEL-ATTRIBUTED EVIDENCE

News tied to Grok 4.6

The model score of 84.8 rates this model's risk profile. The overall Doom Index of 67.9 measures the complete temporally weighted evidence record. These values answer different questions.

NET MODEL-ATTRIBUTED INDEX CONTRIBUTION+0.14

Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.

Autonomy TOWARD

SpaceXAI releases Grok 4.6 for long-running agent work

SpaceXAI released Grok 4.6 through its API and multiple agent platforms, reporting gains over Grok 4.5 on coding, knowledge-work, and long-horizon agent evaluations, plus self-testing and verification during extended tasks.

Full item contribution
+0.14
Grok 4.6 equal share
+0.14
Read assessment →
AUDIT TRAIL

Model score history

  1. R2
    Doom Score 84.8

    The dated official Grok 4.6 release confirms paid API and hosted-platform access. The provider's current enterprise terms (https://x.ai/legal/terms-of-service-enterprise) establish restricted proprietary SaaS access and prohibit reverse engineering. Availability revised from Unknown to Closed. Exact-version evidence chronology replayed under temporal-monthly-pressure-v4.

    14 Aug 2026
  2. R1
    Doom Score 84.8

    Initial profile from the dated Grok 4.6 release, benchmark results, deployment channels, and safeguard disclosure.

    12 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Grok 4.6.

  1. Grok 4.6 by SpaceXAI has a DoomBench model risk score of 84.8 out of 100, based on five transparent version-specific dimensions rather than the overall index.

  2. Grok 4.6's highest current DoomBench dimension is capability at 91.0 out of 100; the profile publishes every component score and its editorial rationale.

  3. DoomBench links 1 source-backed evidence item to Grok 4.6, while keeping the model's risk profile separate from each item's contribution to the live Doom Index.

    https://www.doombench.com/models/spacexai-grok-4-6