GPT-5

Frontier

OpenAI•Released on 2025-08-07

OpenAI's unified flagship model with built-in routing system that auto-selects optimal sub-models. HN users praise its comprehensive multimodal capabilities and competitive pricing ($1.25 vs Claude $15). However, benchmark chart errors at launch sparked controversy.

91
Overall Score

Core Specs

400K
Context Window
128K
Max Output
✓
Reasoning
✗
Open Source
Multimodal Support
textimageaudiovideo

Scenario Scores

Coding
#5 / 886
Writing
#1 / 892
Translation
#1 / 893
Data Analysis
#1 / 891
Conversation
#1 / 894
Image Gen
#1 / 890

Pros & Cons

Pros

  • +Highly competitive pricing
  • +Most comprehensive multimodal
  • +DALL-E integration
  • +Largest ecosystem

Cons

  • −Coding inferior to Claude Opus
  • −Launch quality issues
  • −Black-box architecture

Pricing

Input (per 1M tokens)$1.25
Output (per 1M tokens)$5.00
Subscription$20/month
Free trial available
Updated on 2026-03-06

Available In

Codex CLI
OpenAI
Cursor
Anysphere
Cline
Open Source
GitHub Copilot
GitHub/Microsoft

Compare with Others

GPT-5 vs GPT-5.4 →GPT-5 vs GPT-5.4 Thinking →GPT-5 vs Claude Opus 4.6 →

Benchmarks

sweBench74.9%
mmlu94.6%
humanEval95.1%
aime202594.6%