gpt-5.6-terra

OpenAI's official launch post introduces the GPT-5.6 series as a three-tier family, explicitly naming "Terra" as the balanced model designed for everyday work, positioned between the flagship "Sol" and the fast, affordable "Luna." The post states that Terra delivers competitive performance to GPT-5.5 while being 2x che Regarding capabilities, OpenAI describes GPT-5.6 Sol as the strongest model in the family for agentic work in coding, biology, and cybersecurity, with Terra sharing the family-level architecture as the everyday-work tier. The post emphasizes a strengthened safety stack and additional system-card evaluations but does no

LayerLens's 14 July 2026 benchmark review of GPT-5.6 confirms Terra sits within 3 points of Sol on most benchmarks at half the token cost ($2.50/$15 per 1M tokens versus Sol's $5/$30). On the Artificial Analysis Coding Agent Index, Terra scores 77.4 compared to Sol's 80 and Luna's 74.6, with SWE-Bench Pro showing Terra On context window performance, the review documents that Sol's 1.05M-token context holds at 91.5% on MRCR 8-needle recall in the 256K-512K range, while Luna drops to 41.3% on the same test — though Terra's specific long-context number is not isolated in the excerpt. Sol's "ultra" multi-agent mode (four parallel sub-age

OpenAI's GPT-5.6 family entered general availability on 9 July 2026, thirteen days after the 26 June limited preview. According to the Medium analysis, the GA announcement expanded access for the Sol, Terra, and Luna tiers across ChatGPT, Codex, and the OpenAI API as part of a global rollout, paired with broader evalua For developers evaluating gpt-5.6-terra specifically, the Medium piece documents that Terra sits between Sol and Luna on the capability curve as the balanced tier. The article clarifies that several facts from the June preview — including pricing, API specifications, and safety posture — did not change at GA; the prima

The llm-stats benchmark aggregator page for GPT-5.6 Terra ranks the model at position 16 on the composite LLM Stats Score, with strong capability placement in Coding (8 of 267) and Search (8 of 90), and category standings in Reasoning (16 of 363) and Healthcare (21 of 244). The page lists a blended price of $2.48 per m Specific benchmark scores cited include a perfect 1.00/1 pass rate on the Connectors production benchmark (rank 2, sourced to deploymentsafety.openai.com) — measuring reliable use of connector-based tools in agentic workflows — and a 0.95/1 on HealthBench Consensus (rank 3, length-adjusted) for healthcare performance o

CodeRabbit's engineering benchmark review (July 9, 2026) assesses where GPT-5.6 Sol and Terra fit in coding-agent and code-review workflows. Sol is recommended for teams coming from GPT-5.5 who need a model that follows through on messy repo tasks, long checklists, and multi-file fixes, while Terra is positioned as a c The review notes that Sol does not erase other frontier models — Anthropic's Fable 5 still feels stronger for architectural judgment, and Sonnet 5 retains cleaner comment quality in some review flows — but Sol wins on practical follow-through. The post also discusses the importance of measuring cost per solved task rat

On 26 June 2026, OpenAI announced GPT-5.6 Sol, Terra, and Luna in a limited preview through Codex and the API for a small group of trusted partners, at the request of the US government and without a public waitlist. According to Thurrott's coverage, GPT-5.6 Sol was positioned as OpenAI's most capable model for cybersec For the GPT-5.6 Terra tier specifically, Thurrott reports it offers a GPT-5.5 level of performance at 2x lower cost, positioning it as the balanced everyday-work model between the flagship Sol and the efficient Luna. GPT-5.6 Sol launched with OpenAI's most robust safety stack to date, including strengthened protections

VentureBeat reports on OpenAI's June 26, 2026 limited-preview unveiling of the GPT-5.6 family, naming Sol, Terra, and Luna and positioning Terra for high-volume business tasks such as customer support, internal tools, and document analysis. Sol and Terra set new high benchmark scores, while Luna performs near GPT-5.5 l The article contextualizes the staggered rollout as following a June 2, 2026 executive order directing federal agencies to develop a process for benchmarking and assessing new AI capabilities before wide release. It also notes the release strategy follows an export control order against Anthropic and the removal of acc