xAIPro TierZero Data Retention

Grok 4.6

Name: ARMES
Availability: InStock
Author: ARMES

Run Grok 4.6

Included on PRO plan

Context Window

500K

Input Token Price

$2.00

per 1,000,000 tokens

Output Token Price

$6.00

per 1,000,000 tokens

Speed Rating

Fast

Verified Frontier Benchmarks

Empirical Evaluation Results

Evaluated across standard and verified harnesses

AA Intelligence Index

reasoning

44.2

AA Coding Index

coding

75.9

AnalystAgent

strategy

41.3%

AutomationBench

agentic

67.0%

AA Briefcase

strategy

1534.1 Elo

Critpt

reasoning

19.7%

EnterpriseOps Gym

agentic

48.3%

Gdp Pdf

strategy

17.2%

AA GDPval 2.0

strategy

1643 Elo

GPQA Diamond

reasoning

93.5%

Humanity's Last Exam (HLE)

reasoning

44.1%

Lcr (1.1)

reasoning

81.0%

Mlcr

reasoning

12.2%

Omniscience

science

29.3

SciCode

science

53.0%

Tau3 Banking

finance

50.7%

Terminal-Bench 2.1

coding

88.0%

Terminal-Bench 4.0

coding

17.2%

Openrouter GPQA Diamond

reasoning

93.3%

Openrouter Tau2 Bench Airline

finance

76.0%

Vals EMB 2.0

finance

62.7%

Vals Vibe Code Bench

coding

76.2%

Vals Code Migration

coding

44.6%

Vals Legal Research

legal

48.1%

FrontierCode 1.1

coding

48.0%

DeepSWE / SWE-bench Pro

coding

65.2%

SimpleQA

voice

48.9%

Chess Puzzles

reasoning

31.0%

Epoch GPQA Diamond

reasoning

94.0%

Context Arena MRCR V2 (8-needle)

agentic

81.5%

Surge Chartography (100-tasks)

reasoning

20.5%

Surge Gdp Pdf (100-tasks)

strategy

17.2%

CursorBench 4.0

coding

41.4%

Vals Index (2)

reasoning

59.2%

Vals Finance Agent

finance

53.7%

Vals Terminal Bench 2.1 (2)

coding

78.3%

Harvey LAB-AA

legal

15.8%

AA Agentic Index

agentic

53.4

TAU-Bench Banking

agentic

43.3%

Briefcase

strategy

1534.1 Elo

Gdpval (2)

strategy

1643 Elo

TAU-3 Banking

agentic

43.3%

TAU-2 Airline

agentic

76.0%

DeepSWE v1.1

coding

66.7%

Briefcase v1.1

strategy

1545.1 Elo

GDPval-AA v2.1

strategy

1663.1 Elo

Coding Agent Index (1.5)

agentic

47.0%

Frontiermath Tiers 1 3 (v2)

math

66.0%

FrontierMath Tier 4

math

31.7%

Kernelbench CUDA Glm52 Fused Moe

coding

9.4

Kernelbench CUDA Megaqwen Decode

coding

5.4

Mystery Game Puzzles

reasoning

34.0%

Ebr Bench

reasoning

30.5%

Frontierswe (2)

coding

25.3%

Rsi Exam (0.1)

reasoning

0.4

AA-LCR v1.1

reasoning

81.0%

AA-Omniscience Index

reasoning

29.3

SimpleQA Verified

voice

48.9%

Furniture Assembly

finance

40.0%

Lmarena Agent

agentic

1.2%

Architectural Profile & Capabilities

AA Intelligence Index: 61 (matches GPT-5.6 Sol), GDPval-AA v2: 1753 Elo, CursorBench v3.2: 69.9%, DeepSWE v1.1: 65.9%, FrontierCode v1.1 Extended: 61.3%, APEX-Agents: 57.5%, Terminal-Bench v3.0: 26%, APEX-SWE: 56.4%, AA-Briefcase: 1577 Elo, Harvey LAB: 15.8%, SWE-bench Verified (Vals): 95.60%, GPQA Diamond: 94.9%, HLE: 42.9%, AA Coding Index: 76.8, AA Agentic Index: 58.7, AA Non-Hallucination Rate: 65.7% | xAI's current frontier model (released Aug 12, 2026; knowledge cutoff Feb 1, 2026). Builds on Grok 4.5 with a focus on long-running agents and interactive/visual work. Matches GPT-5.6 Sol on the AA Intelligence Index composite (61). **Domain-leading strengths: knowledge work and professional tasks (GDPval-AA v2 #1 at 1753 Elo), legal agentic work (Harvey LAB 15.8% — 6.3× Sol), in-editor coding (CursorBench 69.9%), long-horizon agentic tasks (APEX-Agents 57.5%).** SWE-bench Verified 95.60% (#4). 500K context. $2/$6 per MTok (unchanged from 4.5), doubling past 200K tokens. **Caveats: trails GPT-5.6 Sol on DeepSWE v1.1 (65.9% vs 73%) and Terminal-Bench v3.0 (26% vs 34.6%) — the launch post skipped these losses; 500K context smaller than peers' 1M; all launch benchmarks vendor-reported, no independent replication yet; pricing doubles past 200K prompt tokens.**

Architecture Type: Dense Transformer

Standard dense autoregressive transformer architecture with full-attention mechanisms.

Recommended Workloads & Primary Use Cases

Everyday reasoning

Drafting

Fast problem solving

Operational Caveats & Boundaries

•Extended context pricing applies above 200K tokens: $4/M in, $12/M out.

ARMES Zero Data Retention

Calls to Grok 4.6 are routed through strict Zero Data Retention inference channels. Prompts and outputs are never stored, indexed, or monitored by xAI.

View Privacy Architecture

More Models from xAI

Compare other engines available in this laboratory.

View Roster (5)

Grok 4.7Pro Tier

Grok 4.7 is SpaceXAI's flagship model for coding, agentic tasks, and knowledge work, succeeding Grok 4.6. Available with Zero Data Retention inference on ARMES AI.

500K ctx$2.00/M

Grok 4.5Pro Tier

Artificial Analysis Intelligence Index: 54 (#4 of 168), AA Coding Agent Index: 76 (on par with GPT-5.5), τ³-Banking: 33% (#1 of 28 models), GPQA Diamond: 93%, Terminal-Bench 2.1: 82–83.3%, SWE-Bench Pro: 64.7%, DeepSWE 1.0: 62%, SWE Marathon: 29% (beats Opus 4.8's 26%), ~15,954 output tokens per SWE-Bench Pro task (4.2× more token-efficient than Opus 4.8) | xAI's flagship model for software engineering, agentic execution, and technical reasoning (released Jul 8, 2026).

500K ctx$2.00/M

Grok 4.3Pro Tier

CaseLaw v2: 79.3% (#1, +25 pts over 4.20), CorpFin: #1, τ²-Bench Telecom: 97.7%, GDPval-AA: 1500 ELO, GPQA Diamond: 90.1%, IFBench: 81.0%, SciCode: 47.3%, Coding Index: 41–42.2%, Terminal-Bench Hard: 38.0%, HLE: 35.0%, Artificial Analysis Intelligence Index: 53 | xAI's most capable reasoning model (beta Apr 17, 2026; GA Apr 30, 2026).

1M ctx$1.25/M

Architectural Profile & Capabilities

Architecture Type: Dense Transformer

Standard dense autoregressive transformer architecture with full-attention mechanisms.