OpenAILanguage modelRetired July 2025
GPT-4.5
The giant non-reasoning model: superb taste and EQ, enormous price.
Overall Intelligence
#7
of 12
10.5 pts behind #1
2
races led
70
weeks at #1
59.7
score
- Released
- Feb 27, 2025
- 1.6 yr ago
- Context
- 128K
- 16K max output
- Price
- $75 / $150
- per 1M in/out · $93.75 blended
- Speed
- 25 tok/s
- 1.4s to first token
- Modalities
- text · image
- outputs text
- Weights
- Closed
- API and apps
The 30-second version
A very large non-reasoning model praised for writing and 'vibes', but too expensive to serve. Retired in July 2025.
Key takeaways
- 1Excellent conversational feel.
- 2Showed the limits of scaling pre-training alone.
- 3Short-lived due to cost.
Where it ranks
Current snapshot · illustrative scores · click any race to watch it
GPT-4.5Best on each axis
Overall
Coding
9Coding+11.6 pts6Frontend & UI+50 Elo9Backend & APIs+11.0 pts8Distributed Systems+8.9 pts9Full-Stack Apps+11.2 pts9Debugging & Repair+11.8 pts9Agentic Coding+14.2 pts9Code Review & Security+10.1 pts9Data Science & SQL+13.1 pts9Mobile Apps+10.6 pts9Terminal & DevOps+13.4 pts9Competitive Programming+543
Hard Reasoning
Math
Science
Writing & Language
Multimodal
Tool Use & Agents
Overall Intelligence over time
#1 frontierGPT-4.5V4.1 FlashGLM-5.3
Update history
The distilled version first; open any release for the deeper breakdown.
The GPT-4 family