Grok 4.3 vs Grok Build 0.1

Best cell per row highlighted. Null means undisclosed — never counted as zero.

xAI
Grok 4.3GA
7.7
AI Panel
xAI
Grok Build 0.1GA
6.7
AI Panel

Identity & lifecycle

ProviderxAIxAI
Family / tierGrok 4 · FlagshipGrok Build · Coder
StatusGAGA
Released2026-04-302026-05-20
Knowledge cutoff2025-122024-11

Architecture & context

Context window1M256K
Max output tokens—256K
Input modalitiestext, image, videotext, image
Reasoning modealwaysalways
Open weightsNoNo

Pricing (per Mtok)

Input$1.25$1
Output$2.5$2
Cached input$0.2$0.2
Batch input$1—
Free tierYesNo

Speed

Output speed (tok/s)181.5—
Time to first token (s)19.69—
Latency tierslowslow

Trust & deployment

Trains on inputsNoNo
LicenseProprietaryProprietary
CloudsAzure AI Foundry—
ComplianceSOC2SOC2

AI Panel scoring

Unified score7.76.7
Decision Maker7.56.5
Domain Strategist86.5
Finance Lead8.57.5
Domain Practitioner7.57
Power User7.56.5
Skeptic6.56
Value score8.57.5

Benchmarks

GPQA Diamond90—
SWE-bench Verified—70.8
IFEval81—
TAU-bench98—
LMArena Elo1443—
LMArena Coding Elo1488—

Scores link to their sources. Missing cells mean the vendor hasn't disclosed a result — honesty over padding.