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.

89
Overall Score

Core Specs

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

Scenario Scores

Coding
#9 / 1486
Writing
#2 / 1492
Translation
#2 / 1490
Data Analysis
#5 / 1488
Conversation
#2 / 1491
Image Gen
#2 / 685

User Feedback Highlights

Based on community feedback. Hover to see original reviews.

+ Most comprehensive multimodal− Launch quality issues− Black-box architecture+ Highly competitive pricing− Coding inferior to Claude Opus+ Largest ecosystem+ DALL-E integration
Sentiment:👍 48%😐 28%👎 24%

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

Reliability

SLA99.9%
Incidents (30d)3
与 GPT-5.4 共享基础设施
View Status Page →

Pricing

Input (per 1M tokens)$1.25
Output (per 1M tokens)$10.00
Subscription$20/month
Free trial available
Updated on 2026-03-06

Tools Supporting This Model

Bolt.new
StackBlitz
Codex CLI
OpenAI
Lovable
Lovable
Perplexity
Perplexity AI
Replit
Replit

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%