GPT-6 Sol and Luna Replace GPT-5.6 Sol and Luna — Sol Moves to Premium 6×

September 23, 2026

GPT-6 Sol and GPT-6 Luna are in the model picker, in the seats GPT-5.6 Sol and Luna held. Same 1.05M-token context, 128K max output, tools, Swarm, and complexity-aware reasoning. Sol lands on Premium 6× — half the SOTA 12× band GPT-5.6 Sol used — and stays Auto-eligible; Luna stays on Budget 1× as a pinned pick and Swarm worker. Open a GPT-5.6 Sol or Luna conversation and the next turn runs on GPT-6. You do not re-pin. GPT-5.6 Terra is unchanged.

What you can do

• Pick GPT-6 Sol from Recommended or Closed SOTA — OpenAI's GPT-6 model built for complex coding and agentic workflows, and the everyday frontier for multi-step reasoning. Quality 68, Premium 6×, Auto and Swarm eligible.
• Start from the new-chat card. It says Try out the latest GPT-6 Sol and Opus 5.5 and opens the picker on the GPT-6 Sol row; Luna sits under Fast. Opus 5.5 stays on Recommended.
• Pick GPT-6 Luna under Fast — OpenAI's most efficient GPT-6 model for focused, high-volume work: chat, classification, extraction, and lightweight agent tasks at Budget 1×. Quality 60, and still a Swarm worker. Auto does not route to Luna; pin it when you want it.
• Keep working in an existing GPT-5.6 Sol or Luna thread. The picker shows the GPT-6 model and the next message runs on it. Saved model preferences roll forward the same way.
• Argus heartbeats default to GPT-6 Luna. Accounts on the old default, and heartbeats still pinned to GPT-5.6 Luna, move over. A different pick or Auto is left alone.
• Scale effort the same way. Fast / Medium / Beast and the complexity router still map onto low / medium / high effort and verbosity, with pro mode at the top band.

Where this shows up

You have a GPT-5.6 Sol analysis open from last week. Open it and keep going — the picker says GPT-6 Sol, and the same thread continues at Premium 6× instead of SOTA 12×.

You run high-volume classification or extraction pinned to GPT-5.6 Luna. Your next batch runs on GPT-6 Luna at the same Budget 1×.

You paired Agent WhatsApp and never touched the heartbeat model. The next Argus check-in runs on GPT-6 Luna.

Try it

• Pick GPT-6 Sol, then: "Challenge this market-entry plan, research the missing evidence, and produce a board-ready recommendation."
• Pick GPT-6 Luna, then: "Classify these files by topic, urgency, and owner, then return a concise action list."
• Pin GPT-6 Sol in Beast mode, then: "Take over this half-finished migration: read the repo, figure out what's left, and finish it — flag anything you had to decide on the way."

Heads up

• GPT-6 Sol is on Premium 6×. GPT-5.6 Sol was SOTA 12×, so each Sol token now draws half as much from your plan quota. GPT-6 Luna stays on Budget 1× (GPT-5.6 Luna's band). Your plan quota and token balance do not change.
• Auto does not pick Luna. Auto routes across Workhorse and SOTA models; GPT-6 Luna is a manual pick under Fast and a Swarm worker.
• GPT-5.6 Sol and GPT-5.6 Luna are retired from the picker, Auto, and Swarm. Sessions, saved preferences, and Argus heartbeats pinned to them roll forward to the GPT-6 model. Stored pins update when you open the chat or send the next message.
• GPT-5.6 Terra stays exactly as it was — same seat, Workhorse 3×.
• No harness numbers yet. Can it Run Alfrada? still shows GPT-5.6 Sol and Luna until GPT-6 has a full run.