GPT-5.6 Luna and Terra get cheaper in Cursor's Other Models pool
Cursor's docs now price GPT-5.6 Luna at $0.2 in / $1.2 out per 1M tokens (down 80% from $1/$6) and GPT-5.6 Terra at $2/$12 (down 20% from $2.5/$15) in the Other Models usage pool.
GPT-5.6 Luna: $1 input / $1.25 cache write / $0.1 cache read / $6 output per 1M tokens. GPT-5.6 Terra: $2.5 / $3.125 / $0.25 / $15.
GPT-5.6 Luna: $0.2 input / $0.25 cache write / $0.02 cache read / $1.2 output per 1M tokens. GPT-5.6 Terra: $2 / $2.5 / $0.2 / $12.
Cursor’s Models & Pricing docs pass through third-party model API pricing at cost inside the Other Models usage pool. As of this capture, GPT-5.6 Luna’s per-token rate dropped roughly 80% across input, cache write, cache read, and output, while GPT-5.6 Terra dropped roughly 20% on the same dimensions. Every other model in both the Cursor Models and Other Models pools (Grok 4.5, Composer 2.5, Claude Fable/Opus/Sonnet 5, Gemini 3.1 Pro, Gemini 3.6 Flash, GPT-5.6 Sol) held steady, and plan-level pricing (Hobby, Pro $20, Pro Plus $60, Ultra $200, Teams Standard $40 / Premium $120, Start ₹649 India) was unchanged. Because Cursor bills these pools at the model’s own API rate, the cut flows straight through to customer bills for anyone drawing on GPT-5.6 Luna or Terra from their included or on-demand usage.
Cursor's docs cut GPT-5.6 Luna's per-token rate roughly 80% (input $1 to $0.2, cache write $1.25 to $0.25, cache read $0.1 to $0.02, output $6 to $1.2 per million tokens) and GPT-5.6 Terra's roughly 20% (input $2.5 to $2, output $15 to $12), passing third-party API price cuts straight through to the dollar-denominated Other Models pool. Plan-level pricing was unchanged.