OpenAI has released GPT-6.1 Terra, a new model designed for visual reasoning and computer use tasks, available via API and within ChatGPT Work and Codex. This model offers a significant cost reduction, priced at $1 per million input tokens and $5 per million output tokens, which is half the cost of its predecessor, GPT-6.1 Sol. While GPT-6.1 Terra demonstrates strong performance in visual reasoning benchmarks like MMMU-Pro (Vision) and excels in computer use tasks such as Agents' Last Exam v1, its coding capabilities on benchmarks like Terminal-Bench 4.0 still lag behind competitors such as Claude Sonnet 5.5. AI
IMPACT This release significantly lowers the cost barrier for visual reasoning and automation tasks, potentially accelerating adoption of AI agents in enterprise workflows.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Agents’ Last Exam v1
- ChatGPT Work
- Claude Opus 5.5
- Claude Sonnet 5.5
- Codex
- GPT-5.6 Terra
- GPT-6.1 Sol
- GPT-6.1 Terra
- GPT-6 Astra
- OpenAI
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →