Core capabilities
- Retained GPT-6 long context window
- Image input, structured outputs and Batch API
- Reasoning levels from none through max
- Escalation path to Sol or Astra for hard cases
OpenAI / current
The high-throughput GPT-6 tier serving free and entry consumer plans.
Retains GPT-6 tooling and long context at a much lower unit price, aimed at high-volume and cost-sensitive workloads. GPT-6 Luna serves free and Go consumer tiers and provides the same Responses API surface with a smaller cost envelope.
Use it as the default tier with an automatic Sol escalation for low-confidence results. Enforce schemas and output caps, and measure document-level accuracy rather than fluency.
Large context at low unit price can hide cost; cap output and cache reusable prefixes. Throughput-oriented tiers may have tighter rate limits; verify concurrency before migration.
Specifications and positioning checked against provider documentation.
Open official source↗