Grok 4.20 vs Grok 4.7 vs Grok Build 0.1

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

xAI
Grok 4.20GA
6.9
AI Panel
xAI
Grok 4.7GA
8.2
AI Panel
xAI
Grok Build 0.1GA
6.7
AI Panel

Identity & lifecycle

ProviderxAIxAIxAI
Family / tierGrok 4 · ReasoningGrok 4 · FlagshipGrok Build · Coder
StatusGAGAGA
Released2026-03-102026-09-212026-05-20
Knowledge cutoff2024-112026-052024-11

Architecture & context

Context window1M500K256K
Max output tokens——256K
Input modalitiestext, imagetext, imagetext, image
Reasoning modeoptionalalwaysalways
Open weightsNoNoNo

Pricing (per Mtok)

Input$1.25$2$1
Output$2.5$6$2
Cached input$0.2$0.5$0.2
Batch input$1——
Free tierYesYesNo

Speed

Output speed (tok/s)171.4——
Time to first token (s)13.24——
Latency tiermediummediumslow

Trust & deployment

Trains on inputsNoNoNo
LicenseProprietaryProprietaryProprietary
Clouds———
ComplianceSOC2SOC2SOC2

AI Panel scoring

Unified score6.98.26.7
Decision Maker786.5
Domain Strategist6.58.56.5
Finance Lead6.597.5
Domain Practitioner7.58.57
Power User6.586.5
Skeptic676
Value score78.87.5

Benchmarks

GPQA Diamond78.5——
MATH-50087.3——
SWE-bench Verified——70.8
IFEval81——
TAU-bench93——
LMArena Elo1475——
LMArena Coding Elo1510——

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