Head to head
Gemini 1.0 Ultra vs Grok-2
5×10²⁵Training compute (FLOP)3×10²⁵
$31MTraining cost$32M
Gemini 1.0 Ultra (Google DeepMind) and Grok-2 (xAI) are both frontier AI models. Gemini 1.0 Ultra was published in December 2023 and Grok-2 in August 2024.
Gemini 1.0 Ultra was trained on 5×10²⁵ FLOP, about 1.7x the compute of Grok-2 at 3×10²⁵ 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.
These two models share no benchmark on which both have been scored, so no direct performance comparison is possible here. The specification table below is a comparison of inputs, not of results.
Specifications
Google DeepMind
Organization
xAI
Dec 6, 2023
Published
Aug 13, 2024
5×10²⁵ FLOP
Training compute1.7x
3×10²⁵ FLOP
Google TPU v4
Training hardware
NVIDIA H100 SXM5 80GB
57,000
Chips used
--
2,400 h
Training time
--
$31M
Training cost (2023 USD)1.0x
$32M
38.4 MW
Training power draw
--
API access
Accessibility
API access
No
Open weights
No
United States
Country
United States
Related comparisons
Grok 4 vs Gemini 1.0 UltraGPT-4.5 vs Gemini 1.0 UltraGPT-4.5 vs Grok-2Grok 3 vs Gemini 1.0 UltraLlama 4 Behemoth (preview) vs Gemini 1.0 UltraLlama 4 Behemoth (preview) vs Grok-2Gemini 1.0 Ultra vs Llama Nemotron Ultra 253BGemini 1.0 Ultra vs Llama 3.1-405BGemini 1.0 Ultra vs Claude 3.5 SonnetGemini 1.0 Ultra vs GPT-4 (Jun 2023)Gemini 1.0 Ultra vs GPT-4 (Mar 2023)Gemini 1.0 Ultra vs Nemotron-4 340B
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.