We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Jan 16·5 min read
Introducing ChatGPT Go, now available worldwide
TL;DR
Advanced AI language models were previously limited by high costs and restricted access, creating barriers for global developers. OpenAI released ChatGPT Go, a new tier offering GPT-5.2 Instant (a faster, lighter model variant) with expanded usage quotas and extended conversation memory.
✦ Why It Matters
Engineers can now build AI-powered applications at lower cost with higher throughput and better context retention across conversations.
Key Takeaways
- Evaluate ChatGPT Go for cost-sensitive production deployments requiring fast inference.
- Leverage extended memory to build stateful conversational applications with richer context.
- Test GPT-5.2 Instant as a lighter alternative to full-scale models for latency-critical tasks.
- Assess global availability to serve users in regions previously blocked by pricing or access restrictions.
Full Summary
Prior to this release, access to state-of-the-art language models—AI systems trained to understand and generate human text—was constrained by pricing and availability limitations. OpenAI built ChatGPT Go, a new service tier providing access to GPT-5.2 Instant, a streamlined version of their flagship model optimized for speed and affordability.
The offering includes higher usage limits (the number of requests or tokens—individual words or subwords—users can process monthly) and longer memory capabilities, allowing conversations to retain more context across interactions. By reducing per-request costs and removing geographic restrictions, ChatGPT Go makes advanced AI inference—the computational process of generating responses—economically viable for startups, researchers, and developers in cost-sensitive regions.
This democratization enables broader experimentation with large language models in production systems globally.