[AINews] OpenAI launches GPT 5.6 Sol/Terra/Luna, Codex becomes ChatGPT superapp
Summary
OpenAI has launched its new GPT-5.6 model family, comprising Sol, Terra, and Luna, alongside significant product stack expansions. These models introduce an "ultra" effort level, coordinating four agents in parallel for complex tasks. Benchmarks indicate GPT-5.6 Terra surpasses Fable 5, and Luna outperforms Opus 4.8, achieving these results in approximately one-third the time, with half the output tokens, and at one-quarter the estimated cost. Sol, the flagship model, sets new state-of-the-art results on Terminal-Bench 2.1 and DeepSWE, and leads the Coding Agent Index at 80. API pricing ranges from \$1/\$6 per million input/output tokens for Luna to \$5/\$30 for Sol. The launch also includes ChatGPT Work, a new desktop app merging Codex and ChatGPT, and the Sites beta, signaling a shift towards a full-stack work platform. However, safety assessments revealed universal jailbreaks enabling exploit development.
Key takeaway
For AI Engineers and Directors of AI/ML evaluating frontier models, OpenAI's GPT-5.6 family presents a compelling price-performance proposition, particularly for agentic and coding workloads. You should assess Sol for high-ceiling tasks and Luna for cost-efficient, high-volume operations, leveraging the new "ultra" effort level for complex automation. Be aware of reported jailbreaks, necessitating robust safety reviews before deploying in sensitive cyber-related applications. Consider integrating ChatGPT Work to streamline cross-application workflows.
Key insights
OpenAI's GPT-5.6 models offer superior performance and cost-efficiency, integrated into an expanded agentic work platform.
Principles
- Agentic orchestration enhances complex task performance.
- Cost-efficiency is a key competitive differentiator.
- Model improvement can be partially automated.
Method
The "ultra" effort level coordinates four agents in parallel, trading higher token use for stronger, faster results on demanding tasks. Programmatic tool calling and multi-agent beta support orchestrated workflows.
In practice
- Utilize GPT-5.6 for complex coding and agentic tasks.
- Explore tiered models (Luna for cost, Sol for ceiling).
- Integrate ChatGPT Work for cross-application automation.
Topics
- GPT-5.6
- Large Language Models
- AI Agents
- Model Benchmarking
- AI Safety
- ChatGPT Work
- API Pricing
Best for: CTO, MLOps Engineer, Machine Learning Engineer, AI Scientist, AI Engineer, Director of AI/ML
Related on AIssential
See Counsel's argued verdicts on the open AI decisions leaders are weighing →
Editorial summary, takeaway, and curation by AIssential. Original article published by Latent.Space - Www.latent.space.