GPT-6 Sol vs Luna

GPT-6 Sol vs GPT-6 Luna

  • $2 / $10
  • $0.10 / $0.50
  • Volume
  • Cost

Try Model

Overview

OpenAI’s GPT-6 mid and low tiers. Sol is the $2 / $10 workhorse (same sticker as Claude Sonnet 5.5). Luna is $0.10 / $0.50 for high-volume, focused jobs—about 1/20 of Sol on input. Astra stays the $10 / $50 flagship. API names: `gpt-6-sol` / `gpt-6.1-sol` and `gpt-6-luna` (confirm the snapshot your host bills).

Standard short context (per million tokens)

GPT-6 SolGPT-6 Luna
Input / output$2 / $10$0.10 / $0.50
Cached input$0.10–$0.20 class*$0.01
Cache write~1.25× input$0.125
RoleDefault GPT-6 workHigh-volume / simple
>272K input2× in/cache, 1.5× outSame long-context rule

Capabilities

  • Sol — GPT-6 daily driverCoding and professional work that used to sit on GPT-5.6 Sol, now at half that promo price ($4/$20 → $2/$10 per OpenAI’s launch note).
  • Luna — cheap loopsClassification, extraction, short codegen, high QPS. Don’t use it as your only model for ambiguous architecture.
  • RouterLuna first, Sol if Luna is weak, Astra once if both fail.

Context Window

Both follow OpenAI’s GPT-6 long-context surcharge: prompts over 272K input tokens reprice the full request (2× input/cache, 1.5× output).

  • Stay under 272K unless you budget for it.
  • Luna at long-context rates is still cheap vs Sol; Sol at long-context is $4 / $15 class.

Pricing

OpenAI standard (short context). Pricing tables also list gpt-6.1-sol at $2 / $10. Batch/Flex 50% off; Fast mode 2×. Regional/FedRAMP uplifts may apply.

GPT-6 Sol
$2 in / $10 out · cached in ~$0.10–$0.20 · write ~$2.50
GPT-6 Luna
$0.10 in / $0.50 out · cached in $0.01 · write $0.125
50K in + 2K out
Sol ≈ $0.12 · Luna ≈ $0.006
  • *OpenAI’s public tables show gpt-6.1-sol cached input at $0.10 and cache writes at $2.50; confirm the exact snapshot id on your invoice.
  • Sol vs Luna is ~20× on input. A chatty agent on Sol can cost more than a tight Luna pipeline.

OpenAI pricing

How to Use

  1. Default Sol for product workFeatures, refactors, multi-file edits.
  2. Default Luna for volumeEval sets, ingest, and cheap tool loops.
  3. Watch Fast mode2× on either SKU. Don’t leave it on for Luna batch jobs.

Alternatives