BenchMark'd
No ratings yet
Be the first to review this model
A cost-efficient variant of GPT-5 offering strong reasoning at a fraction of the price. Replaces o4-mini as the go-to model for developers needing capable intelligence with low latency and high throughput. Supports text and image inputs with 128K output tokens.
Released
August 7, 2025
Parameters
Unknown
Context
400K
Pricing
Paid
| Benchmark | Category | Score | Performance |
|---|---|---|---|
MMLU | knowledge | 88.2% | 88 |
HumanEval | coding | 89.5% | 90 |
MATH | reasoning | 82.1% | 82 |
Chatbot Arena ELO | overall | 1350 | 79 |
Last updated: March 15, 2026
Benchmark scores may vary based on evaluation methodology and conditions.