GPT-5.6: Sol, Terra & Luna
Which tier, what it costs, how to leave 5.4
As of 2026-10-07: GPT-5.6 (Sol, Terra, Luna) was released on 2026-07-09, but it is no longer OpenAI's newest generation — OpenAI released GPT-6 Astra on 2026-09-03, GPT-6 Sol and GPT-6 Luna on 2026-09-22 and GPT-6.1 Sol on 2026-09-29, and its models page now says to start with GPT-6 Astra if you are not sure. All three GPT-5.6 tiers remain in OpenAI's model catalog; official prices are $4/$20 for Sol (promotional at least through 2026-11-21) and $2/$12 for Terra, with Luna the cheapest of the three. This page keeps the GPT-5.6 tier, billing and 5.4-migration details.
Claude, GPT and Chinese models on one key, billed per token; check /models for callable models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
As of 2026-10-07: all three GPT-5.6 tiers are still listed in OpenAI's model catalog, but OpenAI's advice for when you are not sure where to start is now GPT-6 Astra, with GPT-6.1 Sol to balance intelligence and cost and GPT-6 Luna for cost-sensitive, high-volume work. On QCode, gpt-5.6-sol and gpt-5.6-terra are still callable, billed by token (per-model rates on /models); QCode does not currently support gpt-6-luna or gpt-5.6-luna (both removed from /models); please use another GPT model, such as gpt-6.1-sol, gpt-6-sol or gpt-5.6-terra. The ChatGPT-login Codex 5.4 cutoff is a separate official line — see the migration notice on this page.
✅ Confirmed facts
What OpenAI's official release and docs actually confirm.
Default: Terra for daily work
Terra at $2/$12 is the default for most coding work. Sol, now at $4/$20 per the 2026-09-12 archive of the official pricing page (promotional at least through 2026-11-21), is for cross-repo and long agents. Luna at $0.20/$1.20 is volume. Sol Ultra is a high-compute mode of Sol, not a separate price tier.
Quota and scope are searched more than scores
Terminal-Bench 2.1: Sol 88.8% / Terra 84.3% / Luna 82.5%. OpenAI’s @thsottiaux, 2026-07-29 (fetched 2026-08-30): OpenAI has not reduced usage on any subscription plans; Sol was using Codex limits faster than expected; usage limits were reset for all ChatGPT Work and Codex users; typical Sol use should last around 18% longer. The scores stay in the comparison table.
Leaving 5.4: ChatGPT-login 08-31; API key is outside this cutoff
Official wording (2026-07-31; rechecked 2026-08-30, official text still “starting August 31”): ChatGPT-signed-in Codex drops GPT-5.4 / 5.4-mini starting 08-31, mapped to terra / luna. API / Codex-key access is outside this retirement. List prices unchanged as of 2026-08-30.
📌 Preview-era questions, answered at GA
The open questions from the preview, now settled by the GA release.
Context window: 1.05M, now official
OpenAI's docs confirm all three tiers share a 1.05M-token API context window and 128K max output (clients like Codex apply a lower session cap, as with GPT-5.5). The 1.5M figure that circulated during the preview was wrong.
GA date: July 9, 2026
'In the coming weeks' turned out to be July 9 — all three tiers opened to everyone via the API and Codex on the same day.
Cerebras 750 tokens/s: confirmed
OpenAI confirmed Sol shipped on Cerebras in July at up to ~750 tokens/s — the preview-era single-source report proved accurate.
ChatGPT availability
The API and Codex are fully open; GPT-5.6's rollout pace across ChatGPT subscription tiers follows OpenAI's official announcements.
Benchmarks and pricing on this page are cross-checked against OpenAI's official release and docs (first checked 2026-07-10; prices and release dates rechecked 2026-10-15). We refresh this page as official information evolves.
From June 26 preview to July 9 GA
The preview started June 26 with about 20 approved partners; on July 9 all three tiers — Sol, Terra and Luna — opened to everyone via the API and Codex. After that (checked 2026-10-07): OpenAI released GPT-6 Astra on 2026-09-03, GPT-6 Sol and GPT-6 Luna on 2026-09-22, and GPT-6.1 Sol on 2026-09-29.
How to use GPT-5.6 on QCode
QCode is a unified AI-coding API platform for developers; as of 2026-10-07, GPT-5.6 Sol and Terra plus the GPT-6 series are callable.
Sol, Terra and the GPT-6 series
As of 2026-10-07, gpt-5.6-sol and gpt-5.6-terra are callable through QCode, as are gpt-6-astra, gpt-6-sol and gpt-6.1-sol — no re-registering accounts or changing your toolchain. QCode does not currently support gpt-6-luna or gpt-5.6-luna (both removed from /models); please use another GPT model, such as gpt-6.1-sol, gpt-6-sol or gpt-5.6-terra.
One key, many models
One QCode key calls Claude, GPT and the main Chinese models: GPT models go through the OpenAI endpoints (Codex CLI, the OpenAI SDK and similar clients), Claude models through the Anthropic endpoint (Claude Code and similar), switchable any time.
Many payment methods, self-service
Top up with Alipay, WeChat Pay, a card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto (multiple coins and networks); activation is self-service and instant after top-up.
GPT-5.6 FAQ
Has GPT-5.6 been released?
Yes. OpenAI previewed GPT-5.6 (Sol, Terra, Luna) on June 26, 2026 and made all three tiers generally available to everyone via the API and Codex on July 9, 2026. As of 2026-10-07 it is no longer OpenAI's newest generation: GPT-6 Astra was released on 2026-09-03, GPT-6 Sol and GPT-6 Luna on 2026-09-22, and GPT-6.1 Sol on 2026-09-29.
What's the difference between Sol, Terra and Luna?
Sol is the flagship for the hardest work; Terra matches GPT-5.5-class quality at roughly half the price; Luna is the fastest and cheapest. Sol Ultra is Sol's high-compute mode, not a separate price tier.
How much does GPT-5.6 cost?
Per million tokens (input / output): Sol $4 / $20 (promo at least through 2026-11-21), Terra $2 / $12, Luna $0.20 / $1.20. Cached reads keep a 90% discount.
Can I use GPT-5.6 on QCode?
Yes, as of 2026-10-07: gpt-5.6-sol and gpt-5.6-terra are callable through QCode — just pick the model in Codex CLI or any client; billing is by token, with per-model rates on /models. QCode does not currently support gpt-6-luna or gpt-5.6-luna (both removed from /models); please use another GPT model — for volume work, gpt-5.6-terra. The same key also calls gpt-6-astra, gpt-6-sol and gpt-6.1-sol.
GPT-5.6 release dates, status and access (checked 2026-10-15)
When was GPT-5.6 released? API, ChatGPT and Codex dates
GPT-5.6 was released on 2026-07-09, and Sol, Terra and Luna all launched the same day. As checked against OpenAI's API changelog on 2026-10-15, a single 2026-07-09 entry released the GPT-5.6 model family: GPT-5.6 Sol for frontier capability, GPT-5.6 Terra for a balance of intelligence and cost, and GPT-5.6 Luna for efficient, high-volume workloads, available from day one on the Responses, Chat Completions and Batch APIs.
In ChatGPT, OpenAI's Help Center model release notes (web.archive.org copy of 2026-09-27) date “Introducing GPT-5.6 Sol in ChatGPT” to 2026-07-09: GPT-5.6 Sol began rolling out that day to eligible paid ChatGPT plans. Free, Go and logged-out users were not included, and availability of the other GPT-5.6 models varies by product and plan.
In Codex, the models page as of 2026-10-15 states that GPT-5.6 Sol, GPT-5.6 Terra and GPT-5.6 Luna remain available during the rollout of the newer models. In Codex CLI, switch with /model or start a task with codex --model followed by the model id.
- 2026-07-09: the OpenAI API releases GPT-5.6 Sol, Terra and Luna together
- 2026-07-09: GPT-5.6 Sol starts rolling out to eligible paid ChatGPT plans
- 2026-07-30: GPT-5.6 Terra becomes 20% cheaper (OpenAI API changelog)
- 2026-08-21: GPT-5.6 Sol moves to $4 input / $20 output per million tokens, a promotional price that runs at least through 2026-11-21
- From 2026-09-03: OpenAI releases GPT-6 Astra, then GPT-6 Sol and GPT-6 Luna (2026-09-22) and GPT-6.1 Sol (2026-09-29)
Is GPT-5.6 still available, and is it being retired?
Yes, it is still available. As of 2026-10-15 GPT-5.6 Sol, Terra and Luna remain in OpenAI's API model catalog, and OpenAI's deprecations page lists no shutdown date for any of them. Instead, it names gpt-5.6-sol and gpt-5.6-terra as recommended replacements for older snapshots such as gpt-5-2025-08-07, gpt-5-mini-2025-08-07 and o3-2025-04-16, which shut down on 2026-12-11.
The same page sets the notice period: generally available models get at least 6 months' notice before retirement, unless safety or compliance concerns require a faster timeline. So even if a GPT-5.6 retirement is announced later, the policy leaves at least half a year to migrate.
What has changed is OpenAI's recommendation. The model catalog now says to start with GPT-6 Astra if you are not sure, to choose GPT-6.1 Sol to balance intelligence and cost, and GPT-6 Luna for cost-sensitive, high-volume workloads. Projects already running on GPT-5.6 do not need to move right away; new projects can evaluate the GPT-6 series directly.
GPT-5.6 specs: context, output limit and the 272K pricing line
According to the official model pages as of 2026-10-15, all three tiers share the same specs: a 1,050,000-token context window, at most 922,000 input tokens per request, 128,000 max output tokens and a 2026-02-16 knowledge cutoff, with text and image input and text output.
reasoning.effort accepts none, low, medium (the default), high, xhigh and max. The gpt-5.6 alias routes requests to gpt-5.6-sol, so code that uses the alias is using Sol.
Watch the 272K line when budgeting. OpenAI's rule is that prompts with more than 272K input tokens are priced at 2x input and 1.5x output for the full request, and cache writes are billed at 1.25x the uncached input rate. Filling the context toward 1M makes a single request far more expensive than a short one.
Using GPT-5.6 on QCode: models, endpoints and payment
As of 2026-10-15, QCode can call gpt-5.6-sol and gpt-5.6-terra, and the same key also calls Claude and the main Chinese models. QCode does not currently support gpt-5.6-luna or gpt-6-luna (both removed from /models); please use another GPT model, such as gpt-6.1-sol, gpt-6-sol or gpt-5.6-terra.
Fill in the endpoints exactly as QCode's docs give them. For Codex CLI, add a custom provider in ~/.codex/config.toml with base_url set to https://api.qcode.cc/openai and wire_api set to responses, then set model to gpt-5.6-terra or gpt-5.6-sol. For clients that use Chat Completions, such as the official OpenAI SDK, set base_url to https://api.qcode.cc/openai/v1. GPT models only work on these two OpenAI endpoints, not on the Anthropic endpoint.
Billed per token; per-model rates are listed on /models. For GPT models, when the input of a single request (including cached tokens) exceeds 272K tokens, the whole request is billed at the long-context rate: input and cached-read prices ×2, output ×1.5, matching OpenAI's official rule. Each model's rates are on /models, in the >272K row under the price.
- Sign up for QCode and top up with Alipay, WeChat Pay, a card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto (multiple coins and networks)
- Buy a plan on qcode.cc/subscribe; this step deducts balance, it does not take a card payment
- Copy your API key (it starts with cr_) from the dashboard and configure Codex CLI or your SDK with the endpoints above
- Check callable models and live rates on /models
More questions on GPT-5.6 release and access
When were GPT-5.6 Terra and Luna released?
On the same day as Sol: 2026-07-09. OpenAI's API changelog released GPT-5.6 Sol, Terra and Luna in a single 2026-07-09 entry; there was no staggered launch.
When did GPT-5.6 come to ChatGPT?
GPT-5.6 Sol began rolling out in ChatGPT on 2026-07-09 to eligible paid plans; Free, Go and logged-out users were not included, according to OpenAI's Help Center model release notes. Availability of Terra and Luna in ChatGPT varies by product and plan, so check the model picker.
How do I switch to GPT-5.6 in Codex CLI?
Type /model in Codex CLI and pick gpt-5.6-sol or gpt-5.6-terra, or start a task with codex --model gpt-5.6-terra. The Codex models page as of 2026-10-15 says all three GPT-5.6 tiers remain available. With a QCode key, set base_url to https://api.qcode.cc/openai and wire_api to responses.
Is GPT-5.7 the next model after GPT-5.6?
No. As of 2026-10-15 neither OpenAI's API changelog nor its model catalog lists a GPT-5.7. After GPT-5.6, OpenAI released the GPT-6 series: GPT-6 Astra (2026-09-03), GPT-6 Sol and GPT-6 Luna (2026-09-22) and GPT-6.1 Sol (2026-09-29).
Will GPT-5.6 Sol stay at $4 / $20?
OpenAI says the promotional price runs at least through 2026-11-21; after that, OpenAI's pricing page is the reference. That is OpenAI's own API price. On QCode you are billed per token; per-model rates are listed on /models.
How can I pay for GPT-5.6 access on QCode?
On the QCode top-up page you can pay with Alipay, WeChat Pay, a card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto (multiple coins and networks). Activation is self-service right after top-up: copy your API key from the dashboard and call gpt-5.6-sol or gpt-5.6-terra. Billed per token; per-model rates are listed on /models.
Use GPT-5.6 today
Sign up for QCode — gpt-5.6-sol and gpt-5.6-terra plus the Claude and Chinese-model lineups, all on one key; Luna is not currently supported, so please use another GPT model.
Related reading
GPT-5.6 vs Claude Opus 4.8
Sol vs Opus 4.8 on Terminal-Bench and pricing, side by side.
2026 AI Model Radar
Live status of every frontier model this year — available, preview or restricted.
Sol vs Terra vs Luna: how to choose
Benchmarks, prices and use cases across the three tiers — pick your default.