GPT-5 is here: a revolution?
OpenAI has unveiled GPT-5, its most advanced language model to date. With performance that far surpasses GPT-4 on all benchmarks, this new model marks a turning point for the industry.
Key numbers
- Reasoning: +40% accuracy on complex logic tasks
- Code: solves 85% of competitive programming problems (vs 67% for GPT-4)
- Multimodal: natively processes text, image, audio, and video
- Context: 1 million token window
Pricing: a barrier?
The GPT-5 API comes at a premium:
- Input: $30/million tokens
- Output: $60/million tokens
That’s roughly 3x more expensive than GPT-4 Turbo, raising questions about economic viability for many startups.
Impact on competition
Google (Gemini), Anthropic (Claude), and Meta (Llama) will need to accelerate. Competition in AI has never been fiercer.
Our analysis
GPT-5 is technically impressive, but the real question remains: who’s going to pay? Companies are increasingly focused on price-to-performance ratios, and open-source models are closing the gap fast.
The number to remember: OpenAI targets $10 billion in revenue for 2025.