Codestral vs GPT-5.2
Side-by-side benchmark comparison across coding, math, reasoning, speed, and pricing.
GPT-5.2 by OpenAI wins on 10 of 14 benchmarks against Codestral by Mistral, which leads on 3. This head-to-head comparison covers coding, math, reasoning, speed, and pricing metrics from the AI Value Index.
Category-by-Category Breakdown
General Intelligence: In general intelligence, GPT-5.2 scores 1475 on Chatbot Arena ELO compared to Codestral's 1260, while GPT-5.2 scores 83.0% on MMLU-Pro compared to Codestral's 68.0%.
Coding: In coding, GPT-5.2 scores 95.0% on HumanEval+ compared to Codestral's 86.0%, while GPT-5.2 scores 80.0% on SWE-bench Verified compared to Codestral's 40.0%, while GPT-5.2 scores 80.0% on LiveCodeBench compared to Codestral's 48.0%.
Math: In math, GPT-5.2 scores 97.0% on MATH compared to Codestral's 72.0%, while GPT-5.2 scores 99.0% on GSM8K compared to Codestral's 88.0%.
Reasoning: In reasoning, GPT-5.2 scores 92.4% on GPQA Diamond compared to Codestral's 44.0%, while GPT-5.2 scores 54.2% on ARC-AGI compared to Codestral's 24.0%.
Context: In context, GPT-5.2 scores 400K on Context Length compared to Codestral's 262K.
Pricing Comparison
Codestral costs $0.30/1M input tokens and $0.90/1M output tokens, while GPT-5.2 costs $1.8/1M input and $14.0/1M output. Codestral is the more affordable option for API usage.
Speed Comparison
Codestral generates output at 90 tok/s compared to GPT-5.2's 90 tok/s, and the time to first token is 280 ms for Codestral versus 380 ms for GPT-5.2. Both models offer similar output speed.
Verdict
For developers prioritizing affordability, Codestral has the edge. For those who value coding and general intelligence and math, GPT-5.2 is the stronger choice.
View Individual Model Pages
Codestral vs GPT-5.2 — FAQ
Which is better, Codestral or GPT-5.2?
GPT-5.2 wins on more benchmarks overall (10 vs 3). However, the best choice depends on your specific needs — each model excels in different areas.
How does Codestral compare to GPT-5.2 for coding?
GPT-5.2 is better for coding, scoring 80.0% on SWE-bench Verified compared to 40.0%. SWE-bench tests real-world software engineering by resolving actual GitHub issues.
Is Codestral cheaper than GPT-5.2?
Yes, Codestral is cheaper. Codestral costs $0.30/1M input and $0.90/1M output tokens. GPT-5.2 costs $1.8/1M input and $14.0/1M output tokens.
Which is faster, Codestral or GPT-5.2?
Both models generate output at the same speed of 90 tok/s.
What benchmarks does the Codestral vs GPT-5.2 comparison cover?
This comparison covers 14 benchmarks including Chatbot Arena ELO, MMLU-Pro, HumanEval+, MATH, SWE-bench Verified, GPQA Diamond, Output Speed, Time to First Token, and more. Metrics span general intelligence, coding, math, reasoning, speed, and cost categories.