GPT-5.6 Pricing Decoded: What Sol, Terra, and Luna Mean for Your Wallet
// OpenAI shipped GPT-5.6 as three tiers - Sol, Terra, and Luna - with prices from $1/$6 to $5/$30 per million tokens. Here is what the new rate card means for ChatGPT Plus, Pro, and API users.
OpenAI shipped GPT-5.6 as three tiers - Sol, Terra, and Luna - with prices from $1/$6 to $5/$30 per million tokens. Here is what the new rate card means for ChatGPT Plus, Pro, and API users.
The short version
OpenAI shipped GPT-5.6 on June 26, not as one model but as three. Sol is the flagship at $5 per million input tokens and $30 output. Terra costs half as much and matches GPT-5.5. Luna is the volume tier at $1 and $6. The pricing looks orderly on paper. The reality is messier.
If you are a ChatGPT Plus subscriber paying $20 a month, you still cannot use any of them. GPT-5.5 is still your default model. Sol, Terra, and Luna are locked behind a government approval process and limited to about 20 partner organizations through the API and Codex.
If you are a ChatGPT Pro subscriber at $200 a month, you will get access first when the models roll out to ChatGPT, but that date has not been announced.
The launch was not normal. The US government asked OpenAI to stagger the release. OpenAI complied. It is the second frontier model intervention in June, after the administration suspended Anthropic's Fable 5 entirely two weeks earlier. source
What actually shipped
GPT-5.6 is a family, not a single model. OpenAI changed its naming scheme. The number (5.6) marks the generation. The name (Sol, Terra, Luna) marks a durable capability tier that can advance on its own schedule. Each tier has its own rate card.
| Tier | Input / 1M tokens | Output / 1M tokens | Positioning |
|---|---|---|---|
| GPT-5.6 Sol | $5.00 | $30.00 | Flagship, deep reasoning, agentic coding, biology, security |
| GPT-5.6 Terra | $2.50 | $15.00 | Balanced, matches GPT-5.5 at half the price |
| GPT-5.6 Luna | $1.00 | $6.00 | Fast, affordable, high-volume work |
Sol also ships with two heavier modes. Max mode pushes single-model reasoning harder. Ultra mode fans the task out to sub-agents. OpenAI has not published separate pricing for Sol Ultra, confirming only that it costs more than base Sol. source
Terra is the most interesting tier for most developers. OpenAI says it delivers competitive performance with GPT-5.5 while costing 2x less. If that holds in production, Terra replaces GPT-5.5 as the default workhorse model for anyone who does not need Sol's peak capability.
Luna is the volume play. At $1 input and $6 output, it competes with models like Gemini 3.5 Flash ($1.50/$9) and DeepSeek V4 Flash ($0.14/$0.28). It is not the cheapest on the market, but it is OpenAI's cheapest and routes simpler workloads off the expensive tiers.
The pricing story behind the sticker
Sol matches GPT-5.5's rate card exactly: $5 input, $30 output. Same price, better model. OpenAI is effectively giving Sol as a free upgrade at the top end, which is unusual for a company that has been gating its best models behind higher subscription tiers.
The real pricing news is at the bottom. Terra at $2.50/$15 undercuts GPT-5.5 by half. Luna at $1/$6 opens a tier OpenAI did not previously have at this quality level. The range from Luna to Sol spans 5x on input and 5x on output.
For comparison against the rest of the market:
| Model | Input / 1M | Output / 1M |
|---|---|---|
| GPT-5.6 Sol | $5.00 | $30.00 |
| GPT-5.6 Terra | $2.50 | $15.00 |
| GPT-5.6 Luna | $1.00 | $6.00 |
| Claude Opus 4.8 | $5.00 | $25.00 |
| Claude Sonnet 4.6 | $3.00 | $15.00 |
| Gemini 3.1 Pro | $2.00 | $12.00 |
| Gemini 3.5 Flash | $1.50 | $9.00 |
| DeepSeek V4 Flash | $0.14 | $0.28 |
Sol slots in just above Claude Opus 4.8 on output pricing. Terra lands in the same range as Claude Sonnet 4.6. Luna is priced above DeepSeek V4 Flash by a wide margin but undercuts most other low-cost tiers from Google and Anthropic.
The new prompt caching rules also affect effective pricing. Cache writes for GPT-5.6 are billed at 1.25x the uncached input rate, while cache reads keep the 90% discount. The 30-minute minimum cache life is longer than previous generations, which benefits workloads with repeated system prompts. source
The consumer impact: who gets what
The pricing story that matters to most people is not the API rate card. It is what happens inside ChatGPT.
ChatGPT Plus ($20/month). You stay on GPT-5.5. OpenAI confirmed that Plus subscribers will not get Sol access at launch. The company has not specified whether Terra or Luna will eventually land on Plus. The pattern so far (GPT-5.5 locked to Pro) suggests the best new models will continue flowing to the $200 tier first.
ChatGPT Pro ($200/month). You will get access to the full GPT-5.6 family first, but not yet. OpenAI says "in the coming weeks" for broader ChatGPT rollout, contingent on government sign-off.
ChatGPT Free. No change. You stay on GPT-4o class models.
The gap between Plus and Pro is widening. When GPT-5.5 launched in April, it went straight to Pro while Plus kept GPT-5.4. Now Sol is here and Plus is still on GPT-5.5. The $180 difference between Plus and Pro is buying earlier access to each new generation, and the gap is compounding.
The government gate
This launch is about the approval process, not the model.
On June 25, Axios reported that the Trump administration asked OpenAI to stagger the GPT-5.6 release. The next day, OpenAI confirmed the request and announced a limited preview for roughly 20 partner organizations, each vetted by the Office of the National Cyber Director and the Office of Science and Technology Policy. source
OpenAI's blog post made the tension explicit. "We don't believe this kind of government access process should become the long-term default," the company wrote. "It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them." source
This follows the same pattern as Anthropic's Fable 5 in April, which was first restricted to select partners, then suspended entirely on June 12 after the administration ordered foreign-national access removed. The difference is that Fable 5 is gone. GPT-5.6 is merely delayed, with a path to wider release.
For consumers, the government gate means no one can buy their way to GPT-5.6 today. Not Plus. Not Pro. Not even most enterprise accounts. The only people running Sol are the ~20 organizations that passed government review. Everyone else is waiting.
Should you care?
The answer depends on what you are doing now.
If you use ChatGPT for writing, research, and everyday tasks. You can ignore GPT-5.6 for now. GPT-5.5 handles those tasks well. Sol's gains are concentrated in coding, biology, and cybersecurity, not general prose or research. Terra and Luna will eventually reach Plus, but there is no reason to upgrade to Pro just for access to models you cannot use yet.
If you are a developer building on the API. Terra is the tier to watch. If it matches GPT-5.5 quality at half the price, it becomes the new default for production routing. The migration path is simple: swap your model string from gpt-5.5 to gpt-5.6-terra and cut your token bill. You cannot do that today because the preview is gated, but planning for it costs nothing.
If you pay for Pro specifically for latest-model access. You are in the same waiting pattern as everyone else. The $200/month has not unlocked GPT-5.6 yet. OpenAI says it will, but has not said when.
If you are evaluating the $20 versus $200 decision. The gap is widening. Plus gets GPT-5.5 at $20. Pro gets the GPT-5.6 family first at $200. If early access to each new generation is worth $180/month to you, Pro still delivers. If you can wait a few months for models to trickle down, Plus covers your needs at one-tenth the price.
What comes next
OpenAI promised general availability "in the coming weeks." That timeline depends on the government review process, which the June 2 executive order said should take up to 30 days. The July 2 deadline is approaching. If the review completes on schedule, wider API access opens first, followed by ChatGPT and Codex rollout.
The Cerebras-hosted version of Sol, targeting 750 tokens per second, is scheduled for July. That is aimed at latency-sensitive production workloads, not consumer chat.
The bigger structural change is the three-tier naming system. OpenAI has signaled that Sol, Terra, and Luna will advance independently. A Sol-class model could ship without waiting for GPT-5.7. That means the pricing bands are now durable. If you learn to route your workloads across Sol, Terra, and Luna today, the same routing logic should work across future generations without reconfiguration.
For consumers, GPT-5.6 is good but unreachable. You cannot get it yet, and most people do not need it. Watch for the government review to clear, then watch for Terra to land on Plus. Those are the two events that change the calculation. Neither has a date.
Sources: OpenAI official announcement (openai.com), The Verge, VentureBeat, DataCamp, ExplainX, Simon Willison, Lushbinary, OpenAI pricing page, and AI Pricing Guru. All pricing current as of June 29, 2026. Access status reflects the limited preview announced June 26.