GPT-5.6 Models Now Available via API

Date: 2026-07-31
Version: OpenClaw 2026.7.1-2

OpenAI’s GPT-5.6 models are now available through the API with a standard API key. We confirmed all three model IDs with live API calls.

Pricing

OpenAI announced significant reductions for the new generation, including an 80% cut for Luna and 20% for Terra.

Input Output
GPT-5.6 Luna $0.20 / 1M $1.20 / 1M
GPT-5.6 Terra $2.00 / 1M $12.00 / 1M
GPT-5.6 Sol $5.00 / 1M $30.00 / 1M

Fast processing doubles these rates. For Luna, fast mode costs $0.40 input and $2.40 output per million tokens.

Adding GPT-5.6 to OpenClaw

OpenClaw now supports GPT-5.6 models, including their 272k context windows and reasoning capabilities. The service_tier setting must be placed under the model entry in agents.defaults.models. Putting it in the provider catalogue is ignored.

Fast mode at high reasoning

Luna was configured with high reasoning and fast processing enabled.

Normal Fast
Average response time 8.1 seconds 4.1 seconds
Observed throughput 54–105 tokens/sec 150–157 tokens/sec

Fast mode roughly halved total response time and produced much more consistent throughput.

For short replies with little or no reasoning, the difference was negligible. Fixed connection and delivery overhead dominated those requests, so paying twice as much provided little benefit.