whichllmmodel
Back to Dashboard

Mistral AI Mistral Large 3 VS xAI Grok 4.5

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

⚖️Our Take

These models use different coding evaluation benchmarks — with Mistral Large 3 evaluated on live-code-bench (good at single-file apps, building games & UIs, and scripting new logic) and Grok 4.5 on swe-bench-pro (excellent at multi-file repositories, autonomous agents, and industrial codebases) — but Grok 4.5 holds a clear reasoning advantage (+7.5% on GPQA Diamond). However, Mistral Large 3 is a massive 4.0x cheaper to run. Choose Grok 4.5 for complex logic and reasoning tasks, or Mistral Large 3 to optimize your budget for high-volume pipelines.
Was this recommendation helpful?
Model Specs

Mistral Large 3

Open SourceAPI Available

Benchmarks & Scores

Coding (live-code-bench)
34.4%

good at single-file apps, building games & UIs, and scripting new logic

Reasoning (gpqa-diamond)
85.5%

graduate-level science QA

Cost & Performance

Cost (per 1M tokens)4.0x cheaper
$0.75Input: $0.50 | Output: $1.50
Context Window
262.14k tokens
Model Specs

Grok 4.5

Benchmarks & Scores

Coding (swe-bench-pro)
64.7%

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

Reasoning (gpqa-diamond)Winner (+7.5%)
93%

graduate-level science QA

Cost & Performance

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

Frequently Asked Questions about Mistral Large 3 vs Grok 4.5 comparison

Mistral Large 3 is cheaper than Grok 4.5. Mistral Large 3 has a blended cost of $0.75/1M tokens, which is about 4.0x cheaper than Grok 4.5 at $3.00/1M tokens.

For coding tasks, Mistral Large 3 scores 34.4% on live-code-bench (good at single-file apps, building games & UIs, and scripting new logic), while Grok 4.5 scores 64.7% on swe-bench-pro (excellent at multi-file repositories, autonomous agents, and industrial codebases).