openai/gpt-5.6
250K context
GPT-5.6 has been released, and the Sol, Terra, and Luna models are fully available through OurToken with one OpenAI-compatible API.
Explore GPT-5.6 models
Model Introduction
OpenAI gpt-5.6
GPT-5.6 has been released, and the Sol, Terra, and Luna models are fully available through OurToken with one OpenAI-compatible API.
The GPT-5.6 family provides three production routes: Sol for maximum capability, Terra for balance, and Luna for low-cost high-volume traffic. Each model supports text and image input, text output, a 250K context window, up to 128K output tokens, function calling, and separate cache pricing.
Why It Looks Great
- Choose Sol for demanding reasoning, coding, and agent workflows.
- Choose Terra for a practical balance of capability, latency, and cost.
- Choose Luna for frequent requests, high throughput, and the lowest token price.
- Use the same OpenAI-compatible Responses API while changing only the model ID.
- Plan repeated workloads with separate cached input and cache write rates.
Key Features
- Models: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna
- Context Window: 250000 tokens
- Max Output: 128000 tokens
- Input: Text and Image
- Output: Text
- Function Calling: Supported
- OurToken Pricing: 20% of official price for each model
Specifications
Choose the Right GPT-5.6 API Model
Compare the three GPT-5.6 routes by workload fit, token price, caching, and production requirements.
GPT-5.6 Sol
The flagship route for difficult reasoning, advanced coding, repository work, tool use, and multi-step agents.
GPT-5.6 Terra
The balanced route for production coding, assistants, analysis, and tools when Sol-level capability is not required for every request.
GPT-5.6 Luna
The lowest-cost route for high-volume assistants, extraction, classification, content processing, and routine coding support.
Shared 250K Context
All three routes support a 250K context window and up to 128K output tokens for substantial requests.
Unified API Integration
Keep one OpenAI-compatible Responses API implementation and switch among the three complete model IDs.
Transparent Family Pricing
Sol, Terra, and Luna each show input, cached input, cache write, output, official, and discounted prices on their dedicated pages.
How to Select a GPT-5.6 Model
Match each workload to Sol, Terra, or Luna, then validate quality, latency, caching, and total request cost.
Classify the Workload
Identify task difficulty, request volume, latency target, context size, output length, and tool requirements.
01Start with Terra
Use Terra as a balanced baseline for common production coding, assistant, analysis, and automation workloads.
02Escalate to Sol
Move difficult reasoning, coding, or multi-step agent requests to Sol when the baseline misses quality targets.
03Route Routine Work to Luna
Use Luna for frequent and cost-sensitive tasks that meet quality requirements without the higher-priced routes.
04Plan Prompt Caching
Measure stable prompt prefixes and include both cached input and cache write rates in cost calculations.
05Review Dedicated Pages
Open each model page for its exact prices, model ID, examples, features, how-to guidance, and FAQ.
06GPT-5.6 Family FAQ
Answers about GPT-5.6 release, availability, Sol vs Terra vs Luna, context, pricing, caching, and API integration.