Documentation
Models
Last updated:
Choosing a model
Choose by task: Astra for complex reasoning and hard coding, Sol for everyday coding and agents, Terra for versatile code and text, and Luna for fast, high-volume simple tasks. Pick an available version from the current list and check its token prices.
Current models
Prices are per 1M tokens, already 91% below official OpenAI rates.
GPT-6.2 Sol
Current- Model ID
gpt-6.2-sol- Input price
- $0.18 / MTok
- Output price
- $0.90 / MTok
GPT-6.1 Sol
Current- Model ID
gpt-6.1-sol- Input price
- $0.18 / MTok
- Output price
- $0.90 / MTok
GPT-6.1 Terra
Current- Model ID
gpt-6.1-terra- Input price
- $0.09 / MTok
- Output price
- $0.45 / MTok
GPT-6 Astra
Current- Model ID
gpt-6-astra- Input price
- $0.90 / MTok
- Output price
- $4.50 / MTok
GPT-6 Luna
Current- Model ID
gpt-6-luna- Input price
- $0.009 / MTok
- Output price
- $0.045 / MTok
GPT-6 Sol
- Model ID
gpt-6-sol- Input price
- $0.18 / MTok
- Output price
- $0.90 / MTok
GPT-5.6 Luna
- Model ID
gpt-5.6-luna- Input price
- $0.018 / MTok
- Output price
- $0.108 / MTok
GPT-5.6 Sol
- Model ID
gpt-5.6-sol- Input price
- $0.36 / MTok
- Output price
- $1.80 / MTok
GPT-5.6 Terra
- Model ID
gpt-5.6-terra- Input price
- $0.18 / MTok
- Output price
- $1.08 / MTok
Which family fits your task?
The family describes the intended role. Available versions and their current prices are listed separately.
Astra
For complex reasoning, difficult code changes, large refactors and long agent workflows with many constraints.
Complex reasoning and hard coding
Sol
A balanced choice for professional work, everyday coding, code review and agents.
Everyday coding and agents
Terra
A versatile family for everyday code and text: feature work, explanations, documentation and editing.
Everyday code and text
Luna
The fastest family for high-volume simple tasks at lower cost: classification, field extraction and short summaries.
Fast, high-volume simple tasks
Model FAQ
Which model is best for coding?
Start with Sol for everyday coding, code review and agents. Use Astra for complex reasoning, large refactors and difficult debugging; Terra for a mix of code and text; and Luna for simple, repetitive tasks. Test the family on your own project before choosing a current version.
What is vibe coding, and which model should I use?
Vibe coding means describing software changes in natural language, then reviewing the generated code. Sol is a useful starting point for everyday work and agents; Astra suits hard planning and refactors, Terra handles code and text, and Luna handles simple repetitive steps.
Which family should I choose to reduce cost?
Luna is the family to consider for lower-cost, high-volume simple tasks such as classification, field extraction and short summaries. Check the live pricing table for the exact rates of each available version.
Can I use several models in one project?
Yes. Use Sol for everyday coding and agents, Astra for hard reasoning and code, Terra for mixed code and text, and Luna for simple bulk tasks. Select an available version of each family; the key and base URL stay the same.