At a Glance
| Vendor | OpenAI |
| Family | GPT-5 series |
| Launched | OpenAI made GPT-5.6 generally available across ChatGPT, Codex, and the API on July 9, 2026, after a limited preview of GPT-5.6 Sol from June 26. It is the first OpenAI release split into three named tiers instead of base, mini, and nano: Sol is the flagship, Terra the balanced mid-tier, and Luna the low-cost tier. GPT-5.6 replaced GPT-5.5 as the default in ChatGPT. An earlier version of this brief dated the launch to early June with a 256K context; both were wrong and are corrected here. |
| Context window | 1,050,000 tokens on all three tiers, with a 128,000-token maximum output. Luna caps input at 922,000 tokens within that window. Requests with more than 272,000 input tokens are billed at 2x input and 1.5x output for the whole request, so the long-context tier is materially pricier than the headline rate. |
| Pricing | Per million tokens on OpenAI's API model pages as of September 2026: Sol $4 input / $20 output ($0.40 cached input), Terra $2 / $12 ($0.20 cached), Luna $0.20 / $1.20 ($0.02 cached). Sol launched at a higher list price and has been cut since, so older cost estimates overstate it. All three share a February 16, 2026 knowledge cutoff. |
| Access channels | OpenAI API (gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna) via Responses, Chat Completions, and Batch; ChatGPT, where Sol was refreshed and Luna expanded to free users in August; Codex; and Microsoft surfaces via Azure OpenAI. |
Notable Benchmarks
Sol scored 80 on the Artificial Analysis Coding Agent Index v1.1 at launch, 2.8 points ahead of Claude Fable 5, and OpenAI emphasized token efficiency on coding tasks over single-turn quality. Terra is positioned as competitive with GPT-5.5 at roughly half the price. Treat vendor efficiency claims as directional until independent runs on your own workload confirm them.
Strengths
A 1.05M-token window on every tier including the cheapest one, a clear three-price ladder for routing work by difficulty, strong agentic coding scores for Sol, and Luna pricing low enough for high-volume classification and monitoring pipelines.
Limitations
The February 2026 knowledge cutoff is older than the July launch suggests, so anything that happened in spring 2026 has to come from retrieval. The 272K-token pricing threshold catches long-context workloads by surprise. Tier names do not map one to one onto the old base, mini, and nano line, which complicates comparisons with GPT-5.5 baselines.
Brand-Visibility Implications
GPT-5.6 is what ChatGPT users now see, and ChatGPT remains the largest single source of AI referral traffic. Two details matter for brands. First, the February 16, 2026 knowledge cutoff means launches, rebrands, and coverage after mid-February are invisible to parametric recall and reach answers only through ChatGPT search retrieval, so crawlability and fresh, citable pages carry that period. Second, the tier split means free and paid users can get answers from different models; baselines should record which tier produced each answer. Re-run ChatGPT brand baselines from July 9 onward rather than carrying GPT-5.5 numbers forward. See GPT-5.5 brief, GPT-6 Astra brief, and ChatGPT visibility.
How Presenc AI Tracks This Model
Presenc AI monitors brand visibility on OpenAI's GPT-5 series as part of continuous multi-platform AI visibility tracking. We sample OpenAI GPT-5.6 across representative prompt sets daily, compare against competitor performance on the same prompts, and flag material mention-rate changes so brand teams can respond quickly when AI representation shifts.