OpenAI documents three GPT-5.6 API models: Sol, Terra, and Luna. Their model IDs are gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna. OpenAI describes Sol as its flagship, Terra as balancing intelligence and cost, and Luna as designed for cost-sensitive, high-volume workloads. These descriptions are useful starting points, not a guarantee that a particular model will perform best on your tasks. (Sol, Terra, Luna)
The model pages list standard API prices per 1 million tokens: Sol: $4 input / $20 output; Terra: $2 input / $12 output; Luna: $0.20 input / $1.20 output. Sol’s listed price is promotional through November 30, 2026; check its model page after that date rather than assuming the same rate continues. (Sol, Terra, Luna)
GPT-5.6 models at a glance
| Model | OpenAI’s stated positioning | API model ID | Listed price per 1M tokens (input / output) |
|---|---|---|---|
| GPT-5.6 Sol | Flagship model | gpt-5.6-sol |
$4 / $20 |
| GPT-5.6 Terra | Balances intelligence and cost | gpt-5.6-terra |
$2 / $12 |
| GPT-5.6 Luna | Cost-sensitive, high-volume workloads | gpt-5.6-luna |
$0.20 / $1.20 |
These are U.S. dollar prices per million tokens. The model pages also list cached-input rates of $0.40 for Sol, $0.20 for Terra, and $0.02 for Luna per million cached input tokens. Sol’s listed standard price is promotional through November 30, 2026. (Sol, Terra, Luna)
Which GPT-5.6 model should you choose?
- Start with Sol when you want to evaluate the flagship option for demanding work. OpenAI’s launch announcement also describes an “ultra” setting for coordinating agents across parallel workstreams. That launch description alone does not establish that Ultra is available as a production API setting, so check the API documentation before designing around it. (Sol model page, GPT-5.6 announcement)
- Try Terra when you want a middle option and cost matters. OpenAI describes it as balancing intelligence and cost. Compare its results with Sol on representative tasks rather than assuming a particular level of equivalence. (Terra model page)
- Try Luna for cost-sensitive, high-volume workloads. OpenAI describes it as designed for that use case. Validate output quality and task completion before routing production traffic to the lowest-priced tier. (Luna model page)
These are starting points based on OpenAI’s model descriptions, not a ranking established by comparable benchmark results. For a production choice, compare accuracy, successful task completion, latency, and total cost on representative examples from your workload.
API pricing: what it costs and what can change the bill
| Model | Input per 1M tokens | Cached input per 1M tokens | Output per 1M tokens |
|---|---|---|---|
| GPT-5.6 Sol | $4.00 | $0.40 | $20.00 |
| GPT-5.6 Terra | $2.00 | $0.20 | $12.00 |
| GPT-5.6 Luna | $0.20 | $0.02 | $1.20 |
Sol-specific long-prompt qualification: its model page says prompts with more than 272,000 input tokens are priced at 2× the input rate and 1.5× the output rate for the full request. The supplied pricing information does not establish that this surcharge applies to Terra or Luna, so do not apply Sol’s rule to those models without checking their current pricing pages. (Sol model page)
Worked example: Suppose your application processes 1 million input tokens and generates 200,000 output tokens in total, and every individual prompt is below 272,000 input tokens. Using the listed standard rates and ignoring cached tokens:
- Sol: (1 × $4) + (0.2 × $20) = $8
- Terra: (1 × $2) + (0.2 × $12) = $4.40
- Luna: (1 × $0.20) + (0.2 × $1.20) = $0.44
This illustrates token billing; it does not show that the models will complete equivalent work equally well or use the same number of tokens. API prices are separate from ChatGPT subscription pricing.
Production API availability and model IDs
OpenAI publishes a model page for each of the three API model IDs. The supplied model-page information establishes a 1,050,000-token context window and 128,000 maximum output tokens for Sol. Those limits should not be generalized to Terra or Luna based on the evidence summarized here; check each model’s current page for its own limits. (Sol, Terra, Luna)
Likewise, verify endpoint support, input modalities, output capabilities, and any API-specific settings on the individual model page before building around them. The specifications summarized here do not establish those details for all three models. A model’s appearance in rate-limit information also does not by itself confirm that your organization can access it; check the Playground or API model list for your account. (Luna model page)
Bottom line
For GPT-5.6 API selection, Sol is the flagship starting point, Terra is positioned to balance capability and cost, and Luna is positioned for cost-sensitive, high-volume work. Their listed prices differ substantially, but the right production choice depends on measured results for your workload, usage patterns, and account access. Apply the greater-than-272,000-token surcharge only to Sol unless the other model pages document the same rule, and recheck Sol’s promotional rate after November 30, 2026.
Sources
GPT-5.6 Sol Model | OpenAI API
GPT-5.6 Terra Model | OpenAI API
GPT-5.6 Luna Model | OpenAI API
GPT-5.6: Frontier intelligence that scales with your ambition | OpenAI