ÆAAI
← All models
GRK420xAI · Grok 4Synthesized

Grok 4.20

xAI's reasoning flagship — AA Intelligence Index 49, 2M context, $2/$6 per MTok.

Accounting overall

66.0%

Input / Output

$2.00 / $6.00 per MTok

Context

2M

Speed

~185 tok/s

Released

2026-03

Cutoff

2026-02

01Accounting Task Breakdown

Eight accounting-task categories borrowed from DualEntry's 101-task benchmark. Measured where published, synthesized from adjacent benchmarks otherwise.

Transaction Class.
80.0
Journal Entry
78.0
Accounts Payable
70.0
Accounts Receivable
68.0
Bank Reconciliation
64.0
Financial Reporting
52.0
Month-End Close
40.0
Accounting Knowledge
76.0
02Research

Grok 4.20 launched as public beta February 17, 2026, with the general reasoning endpoint going live March 31, 2026. ~1T parameters, 2M context. Artificial Analysis Intelligence Index of 49 (reasoning variant) ranks #15 of 135 — meaningfully above Grok 4.1 Fast (AA Index 24). Reported SWE-bench 75%, first model to break 10% on ARC-AGI at 15.9%.

Accounting sub-scores are **synthesized** — not yet on DualEntry. We estimate ~69% overall, placing it between GLM-5 (72.3%) and MiniMax M2.7 (71.3%) on structured tasks and somewhat below on judgment-heavy close work. The $2/$6 pricing positions Grok 4.20 as a Gemini 3.1 Pro alternative rather than a budget option — a premium tier alongside, not a replacement for, Grok 4.1 Fast.

For buyers currently running Grok 4.1 Fast, Grok 4.20 is the capability-upgrade path; for those chasing absolute cost-efficiency, Grok 4.1 Fast at $0.20/$0.50 stays the correct choice. Both deserve entries.

Citations