SShogo Bench

GPT-5.4 vs Grok 4.7

Grok 4.7 leads GPT-5.4 by 3.2 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. GPT-5.4 wins 9 and Grok 4.7 wins 8 of 17 shared benchmarks. Grok 4.7 is cheaper for agent workloads.

Add more models to this comparison →

Pillars

PillarGPT-5.4Grok 4.7
Agentic coding5758
Long-horizon autonomy6169
Knowledge-work agents4047
Computer use4239
Security & context learning5450

Shared benchmarks

BenchmarkGPT-5.4Grok 4.7
APEX-Agentsindex52.4%70.0%
Vending-Bench 2index$6,144$10,537
ARC-AGI93.7%90.2%
ARC-AGI-274.0%61.4%
Chess Puzzles48.0%41.0%
CritPt23.4%18.0%
DTBench94.4%96.0%
FrontierMath Tier 449.0%17.1%
FrontierMath Tiers 1-378.6%53.0%
GPQA Diamond93.3%92.7%
LMCA52.0%49.4%
MirrorCode23.3%54.4%
OTIS Mock AIME97.8%98.1%
ProofBench56.0%26.0%
SciCode56.6%57.8%
SimpleQA Verified45.1%56.0%
WebDev Arena14631638

Specs and price

GPT-5.4Grok 4.7
Agent Index51.6 (#33)54.8 (#27)
Agent $/M$1.27$0.98
Input / output $/M$2.50 / $15.00$2.00 / $6.00
Context1.1M500K
ReleasedMar 5, 2026Sep 21, 2026
WeightsClosedClosed
API hosts45

Don’t pick one. Route.

Shogo can run GPT-5.4 and Grok 4.7 side by side and send each step of an agent to the one that’s cheaper for it.

Try both in Shogo