Ao vivo
Head to head

GPT-4 (Jun 2023) vs Grok 3

OpenAI
GPT-4 (Jun 2023)
June 2023
vs
xAI
Grok 3
February 2025
2.1×10²⁵Training compute (FLOP)3.5×10²⁶
121.9Capability index139
--Training cost$218M
Benchmarks won (1 scored on both)
0 GPT-4 (Jun 2023)Grok 3 1

GPT-4 (Jun 2023) (OpenAI) and Grok 3 (xAI) are both frontier AI models. GPT-4 (Jun 2023) was published in June 2023 and Grok 3 in February 2025.

Grok 3 was trained on 3.5×10²⁶ FLOP, about 16.7x the compute of GPT-4 (Jun 2023) at 2.1×10²⁵ FLOP. Training compute is the closest available proxy for how much was invested in a model, though it says nothing on its own about how well that compute was spent.

Grok 3 wins the single benchmark both were scored on. Benchmark counts are a crude scoreboard — the evaluations differ wildly in what they measure and in how saturated they are — so the per-benchmark table below matters more than the tally.

Specifications
OpenAI
Organization
xAI
Jun 13, 2023
Published
Feb 17, 2025
2.1×10²⁵ FLOP
Training compute17x
3.5×10²⁶ FLOP
1.8T
Parameters1.7x
3T
5.4T
Dataset size
--
NVIDIA A100 SXM4 40 GB
Training hardware
NVIDIA H100 SXM5 80GB
25,000
Chips used3.2x
80,000
2,280 h
Training time
2,160 h
--
Training cost (2023 USD)
$218M
19.9 MW
Training power draw
109.9 MW
API access
Accessibility
API access
No
Open weights
No
United States
Country
United States
Shared benchmarks
Related comparisons
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.