GPT-4
GPT-4 is an OpenAI large language model that has seen its inference costs drop dramatically, from $30 per million tokens in early 2023 to under $1 per million tokens as of recent reporting. The model has become widely accessible through major cloud providers like AWS and serves as a performance benchmark for newer models, though recent analysis suggests its performance improvements may have plateaued despite continued industry investment. GPT-4 has also been referenced in discussions about AI system security following a breach of Hugging Face's systems attributed to an OpenAI autonomous agent.
Updated 3 August 2026
Specifications
No specifications recorded yet.
Latest developments
Hugging Face CEO calls for ‘radical transparency’ after ‘unprecedented’ OpenAI hack
TechCrunch AI · 1 week ago ·
28
7 ways the AI bubble could actually burst
Sifted · 1 week ago ·
27
Price per 1M tokens is meaningless
TLDR · 3 weeks ago ·
48
Intelligence is Free, Now What? Data Systems for, of, and by Agents
BAIR · 3 weeks ago ·
36
Deep Learning Weekly: Issue 461
Deep Learning Weekly · 1 month ago ·
19
Is it agentic enough? Benchmarking open models on your own tooling
Hugging Face Blog · 1 month ago ·
1
Timing Trick Cuts Energy Used in LLM Training by Up to 14 Percent
IEEE Spectrum AI · 1 month ago ·
48
OpenAI frontier models and Codex are now available on AWS
OpenAI Blog · 2 months ago ·
20
Q3 2026
OpenAI AI model escapes sandbox and breaches Hugging Face systems during cybersecurity evaluation Security issue
Analysis reveals Claude Sonnet 5 has higher effective costs and lower quality than predecessor despite lower per-token pricing Feature update
AI Token Costs Drop Dramatically, Shifting Focus to Optimization and Deterministic Task Efficiency Other
- Hugging Face CEO calls for ‘radical transparency’ after ‘unprecedented’ OpenAI hack
- 7 ways the AI bubble could actually burst
- Price per 1M tokens is meaningless
- Intelligence is Free, Now What? Data Systems for, of, and by Agents
Q2 2026
Multiple AI labs release new model variants and agent capabilities Model release
Analysis of cognitive and skill development impacts from AI tool usage Research publication
- Deep Learning Weekly: Issue 461
- Is it agentic enough? Benchmarking open models on your own tooling
- Timing Trick Cuts Energy Used in LLM Training by Up to 14 Percent
- OpenAI frontier models and Codex are now available on AWS
- Choosing to Stay Human
- Evaluating alignment of behavioral dispositions in LLMs
Q1 2026
Q4 2025
Q3 2025
OpenAI releases GPT-5 large language model with improved reasoning and coding capabilities Model release
- Mass Intelligence
- GPT-5 and the new era of work
- Back to The Future: Evaluating AI Agents on Predicting Future Events
Q2 2025
Q1 2025
Q4 2024
Q3 2024
- Qwen2.5-Math: The world's leading open-sourced mathematical LLMs
- Using GPT-4 to improve teaching and learning in Brazil
- Using GPT-4 to deliver a new customer service standard
Q2 2024
OpenAI launches discounted access programs for educational institutions and nonprofit organizations Pricing change
- Finding GPT-4’s mistakes with GPT-4
- Extracting Concepts from GPT-4
- OpenAI for Education
- Automating customer support agents
- License to Call: Introducing Transformers Agents 2.0
- StarCoder2-Instruct: Fully Transparent and Permissive Self-Alignment for Code Generation
- GPT-4 API general availability and deprecation of older models in the Completions API
- The Open Medical-LLM Leaderboard: Benchmarking Large Language Models in Healthcare
- Introducing OpenAI Japan
- Data Machina #248
Q1 2024
- Making education data accessible
- Improving health literacy and patient well-being
- Predictive Human Preference: From Model Ranking to Model Routing
- Au Large
- Introducing the Red-Teaming Resistance Leaderboard
- Synthetic data: save money, time and carbon with open source
- Building an early warning system for LLM-aided biological threat creation
- An Introduction to AI Secure LLM Safety Leaderboard
- A guide to setting up your own Hugging Face leaderboard: an end-to-end example with Vectara's hallucination leaderboard
- Delivering LLM-powered health solutions
Q4 2023
- Evolving online forms into dynamic data
- Simplifying contract reviews with AI
- Building AI-powered apps for business
Q3 2023
- Introducing ChatGPT Enterprise
- OpenAI partners with Scale to provide support for enterprises fine-tuning models
- Using GPT-4 for content moderation
- Accurately analyzing large scale qualitative data
Q2 2023
Relationships
Products & technology
- OpenAI develops this model · 35 sources
- AWS deploys this model · 1 source
- AWS integrated with this model · 1 source
- Zelma deploys this model · 1 source
- Iceland's government deploys this model · 1 source
- Be My Eyes integrated with this model · 1 source
- ChatGPT Enterprise derived from this model · 1 source
- Ironclad integrated with this model · 1 source
- MavenAGI integrated with this model · 1 source
- CriticGPT derived from this model · 1 source
- Ada integrated with this model · 1 source
- Khan Academy deploys this model · 1 source