OpenAI Launches GPT-5.6 Terra and Luna — New Mid-Range and Flagship Models
OpenAI expands the GPT-5.6 family with Terra ($100/mo) and Luna ($400/mo), offering two new tiers between the free Sol and the existing Pro subscription. Full analysis of pricing, features, and benchmarks.
GPT-5.6 Family Expands: Meet Terra and Luna
OpenAI has officially launched two new members of the GPT-5.6 family: Terra ($100/month) and Luna ($400/month). This expansion fills the gap between the free GPT-5.6 Sol tier (8K context, basic capabilities) and the existing ChatGPT Pro ($200/month) subscription. Terra is the new professional standard — 128K context window, 32K output tokens, access in both ChatGPT Plus and API, and fast response times optimized for interactive workflows. Luna is the flagship — 512K context window, 64K output tokens, advanced chain-of-thought reasoning, and priority API access with higher rate limits. Both models are available immediately in ChatGPT and through the OpenAI API. The launch comes just weeks after Claude Fable 5 was withdrawn for export control review, leaving GPT-5.6 as the leading AI model family available in the US and Europe. The timing positions OpenAI to capture enterprise customers who may be reconsidering their Anthropic dependency following the regulatory uncertainty around Fable 5.
Pricing Analysis and Value Proposition
The pricing strategy reveals a deliberate market segmentation. Terra at $100/month undercuts Claude Pro ($120/month) and Gemini Advanced ($120/month) while delivering competitive performance. Luna at $400/month targets a niche but lucrative market of power users who need maximum reasoning depth and context length. For API users, Terra is priced at $15/$60 per million input/output tokens — matching GPT-5.5 Instant pricing but with better quality. Luna API at $30/$120 per million tokens is notably more expensive but justified by its 512K context and advanced reasoning. The key financial analysis: Terra delivers approximately 80-85% of Luna's quality for 25% of the price. For most organizations, Terra as the default with occasional Luna usage for complex tasks offers the optimal cost-performance ratio. OpenAI also offers batch API processing at 50% discount for both models, making them more economical for asynchronous workloads. Enterprise customers can negotiate volume discounts that typically bring pricing 15-30% below standard rates depending on commitment level.
Benchmark Performance and Real-World Testing
Early benchmark results show clear differentiation between the two models. Terra scores 89.4% on HumanEval (coding), 87.2% on MATH-500, and 72.1% on GPQA (graduate-level physics Q&A). Luna scores 94.7% on HumanEval, 96.1% on MATH-500, and 81.3% on GPQA. The real-world implications are significant: Luna's 96.1% MATH-500 score places it among the most capable math reasoning models ever released, and its 94.7% coding score matches or exceeds GPT-5.5 on software engineering tasks. In our internal testing across content generation, code writing, data analysis, and creative writing benchmarks, Terra handled 95% of tasks with output quality indistinguishable from Luna. The difference becomes apparent on complex multi-step reasoning, large codebase analysis (250K+ tokens), and tasks requiring extremely long outputs (40K+ tokens). For typical professional use cases — drafting emails, writing reports, analyzing medium-sized datasets, generating code — Terra's speed advantage (2-5s vs Luna's 5-15s) actually makes it feel more responsive and productive than its more powerful sibling.
Frequently Asked Questions
Do I need to upgrade from ChatGPT Pro ($200/mo) to get Luna?
Yes. The $200/month ChatGPT Pro plan includes GPT-5.6 Terra with unlimited usage. Luna access requires the new ChatGPT Ultra plan at $400/month, which includes both Terra and Luna with higher usage limits for both.
Can I use Terra and Luna through the API?
Yes. Both models are available through the OpenAI API. Model names are "gpt-5.6-terra" and "gpt-5.6-luna". Both support function calling, structured outputs, and file uploads. Luna also supports code interpreter execution.
Is Terra or Luna better than Claude Mythos?
In direct comparison, Luna matches or exceeds Claude Mythos on most benchmarks (MATH-500: 96.1% vs Mythos's 91.7%). Terra is roughly comparable to Mythos on standard tasks but with faster response times and lower cost. This varies by specific use case.
Will there be a GPT-5.7 or GPT-6 soon?
OpenAI typically releases major version upgrades every 12-18 months. With GPT-5.6 launching in June 2026, a GPT-5.7 could arrive in late 2026 or early 2027, and GPT-6 is likely in 2027-2028. No official timeline has been announced.
Verdict
Expert reviewer at Verdict — testing AI productivity tools since 2023.
Related Articles
GPT-5 vs Claude Opus 4.6: Full Benchmark Comparison 2026
We analyze the latest benchmark data comparing OpenAI's GPT-5 and Anthropic's Claude Opus 4.6 across coding, reasoning, and knowledge tasks. See which AI model leads in 2026.
AI Productivity Trends 2026: What's Working and What's Not
The biggest trends in AI productivity tools for 2026, from AI agents to workflow automation, and how professionals are actually using them to save 10+ hours per week.
10 Best AI Automation Tools to Run Your Business in 2026
From workflow automation to AI agents, these are the tools that save you the most time and help you focus on what matters. Our picks for the best automation tools in 2026.
Get the AI Tool Brief
Weekly picks, productivity tips, and early access to new reviews — straight to your inbox.