whichllmmodel
Back to Dashboard

xAI Grok 4.20 VS xAI Grok 4.5

✍️ Analysis by:the whichllmmodel Editorial Team|📅 Updated: June 2026

👑Our Take

Grok 4.5 is the superior choice here. It holds a 8.0% average lead in benchmarks while actually being more cost-effective (saving 1.0x on API costs compared to Grok 4.20). Unless you have specific provider lock-in, go with Grok 4.5.
Was this recommendation helpful?
Model Specs

Grok 4.20

Benchmarks & Scores

Coding (swe-bench-pro)
51.8%

excellent at multi-file repositories, autonomous agents, and industrial codebases

Reasoning (gpqa-diamond)
90%

graduate-level science QA

Cost & Performance

Cost (per 1M tokens)
$3.00Input: $2.00 | Output: $6.00
Context WindowLarger
1.05M tokens
Model Specs

Grok 4.5

Benchmarks & Scores

Coding (swe-bench-pro)Winner (+12.9%)
64.7%

excellent at multi-file repositories, autonomous agents, and industrial codebases

Reasoning (gpqa-diamond)Winner (+3.0%)
93%

graduate-level science QA

Cost & Performance

Cost (per 1M tokens)
$3.00Input: $2.00 | Output: $6.00
Context Window
500k tokens

Frequently Asked Questions about Grok 4.20 vs Grok 4.5 comparison

Both Grok 4.20 and Grok 4.5 cost the same, with a blended cost of $3.00 per 1 million tokens.

Grok 4.5 is better for coding tasks on this benchmark. It scores 64.7% on swe-bench-pro (excellent at multi-file repositories, autonomous agents, and industrial codebases) compared to Grok 4.20 which scores 51.8%.