OpenAI Expands GPT-5.6 Access Across ChatGPT, Codex, and API

GPT-5.6 availability is expanding as OpenAI rolls out its full GPT-5.6 model family across ChatGPT, Codex, and the OpenAI API.

The new lineup includes GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna. The models were previously available in a limited preview and are expected to become available over the next 24 hours.

GPT-5.6 Sol is the flagship model

GPT-5.6 Sol is the most capable model in the new family. OpenAI designed it for complex reasoning, coding, scientific research, cybersecurity, and long-running agentic tasks.

The model supports a new max reasoning effort setting for harder problems that need more computation. It also introduces ultra mode, which uses multiple subagents to speed up complex reasoning workflows.

OpenAI will also offer a faster version of GPT-5.6 Sol through Cerebras, with speeds of up to 750 tokens per second.

GPT-5.6 Terra targets balanced performance

GPT-5.6 Terra is the balanced model in the lineup. OpenAI says it delivers performance competitive with GPT-5.5 at about half the price.

The model targets users and developers who need strong performance without the higher cost of Sol. ChatGPT Free and Go users will receive access to GPT-5.6 Terra.

GPT-5.6 Luna is the low-cost option

GPT-5.6 Luna is the smallest and most affordable model in the GPT-5.6 family.

It is designed for fast, lower-cost workloads where users do not need the full reasoning power of Sol. Plus, Pro, Business, and Enterprise users can choose Luna alongside Sol and Terra.

GPT-5.6 adds stronger reasoning and agent features

GPT-5.6 Sol introduces max reasoning effort for tasks that benefit from deeper analysis. Ultra mode goes further by delegating work to subagents instead of relying on a single-agent workflow.

Max mode will be available to users with GPT-5.6 access in ChatGPT Work and Codex. ChatGPT Work users on Pro and Enterprise plans can access GPT-5.6 Sol ultra, while Codex users on Plus and higher plans can also use ultra mode.

GPT-5.6 Sol improves benchmark performance

OpenAI says GPT-5.6 Sol scored 53.6 on Agents’ Last Exam, beating Claude Fable 5 by 13.1 points. At medium reasoning effort, Sol still led by 11.4 points while using roughly one-quarter of the estimated cost.

On the Artificial Analysis Intelligence Index, Sol came within one point of Fable 5. OpenAI says it completed those tasks 61% faster at around half the estimated cost.

On the Coding Agent Index, GPT-5.6 Sol scored 80, beating Fable 5 by 2.8 points. It reportedly used less than half the output tokens and time, with about one-third lower estimated cost.

Sol also surpassed Claude Opus 4.8 on OSWorld 2.0 while using 85% fewer output tokens.

GPT-5.6 Sol brings cybersecurity gains

GPT-5.6 Sol also shows major gains in cybersecurity benchmarks.

The model scored 73.5% on ExploitBench 1, compared with 47.9% for GPT-5.5 at a similar output-token budget.

On ExploitGym 2, Sol reached a 24.9% pass rate within two hours, up from GPT-5.5’s 15.1%. With a six-hour limit, the score increased to 33.7%.

On SEC-Bench Pro, GPT-5.6 Sol scored 71.2%, compared with 45.8% for GPT-5.5. OpenAI also reported lower latency for Sol on cybersecurity tasks.

GPT-5.6 API pricing

GPT-5.6 Sol costs $5 per million input tokens and $30 per million output tokens, while GPT-5.6 Terra costs $2.50 per million input tokens and $15 per million output tokens.

GPT-5.6 Luna is the most affordable option, priced at $1 per million input tokens and $6 per million output tokens.

Who gets GPT-5.6 in ChatGPT?

ChatGPT Plus, Pro, Business, and Enterprise users can access GPT-5.6 Sol with medium and higher reasoning settings.

Pro and Enterprise users can select GPT-5.6 Sol Pro for especially complex tasks. Free and Go users will get access to GPT-5.6 Terra.

Plus, Pro, Business, and Enterprise users can choose between Sol, Terra, and Luna. Eligible users can also select a reasoning effort level for each model.

The rollout gives OpenAI a broader GPT-5.6 lineup for different needs, from high-end reasoning and agentic workflows to lower-cost everyday tasks.