OpenAI launches GPT-5.6 family Luna Terra Sol with new per-token pricing
TL;DR
OpenAI released the GPT-5.6 model family in three sizes with pricing of $1/$6 for Luna, $2.50/$15 for Terra, and $5/$30 for Sol per million input/output tokens.
What changed
OpenAI released the GPT-5.6 family today with Luna, Terra, and Sol sizes now in general availability. Luna starts at one dollar per million input tokens and six dollars output while Terra and Sol scale up to five dollars input and thirty dollars output. Vibe Builders, Basic Users, and Developers can pick the size that matches their needs.
Why it matters
The Claude Opus series lists at five dollars input and twenty five dollars output for the same token measure. Developers gain clear options to align costs with task scale while Vibe Builders match model size to creative workflows. Basic Users benefit from the tiered structure that keeps smaller runs affordable.
What to watch for
Compare against the Claude Fable 5 at ten dollars input and fifty dollars output on identical prompts. Developers should run a fixed task on both GPT-5.6 Sol and Claude Opus then inspect the actual token counts reported in the API response.
Who this matters for
- Vibe Builders: Use the cheap Luna model for rapid prompt drafting before scaling up to Sol for final assets.
Harsh’s take
The GPT-5.6 pricing tier shows OpenAI is targeting different operational budgets directly. Luna at $1/$6 makes high-volume testing cheap, while Sol at $5/$30 challenges Anthropic's high-end pricing. Do not rely on raw token prices to estimate your actual run costs. Because reasoning models generate variable internal tokens, you must run identical test prompts on both Sol and Claude Opus to compare the final API bills.
by Harsh Desai
More AI news
- Daily RoundupReplicate voice clone and turbo image model, Ornith-1.5 trends, Ox Alpha speculation
Replicate added a zero-shot music voice cloning tool and a fast 6B-parameter image LoRA while Ornith-1.5-9B gained traction on Hugging Face and a stealth model called Ox Alpha stirred online talk.
- Daily RoundupQwen3.8 models trend on Hugging Face, voice-clone-pro on Replicate, and CPU-first agents
On 22 August new Qwen variants and small CPU models hit Hugging Face while voice cloning and image-to-video tools launched on Replicate and Fal, alongside fresh analytics and port tools plus updates on AI safety and research agents.