Skip to main content
AI Choice Engine

Model comparison

GPT-5.6 Sol vs GPT-6 Luna

Two OpenAI tiers compared on the figures that decide which one a workload actually needs.

Catalog record checked September 26, 2026Individual provider fields may changeEvidence confidence: High — see receipts below

OpenAI

GPT-5.6 Sol

Frontier

vs

OpenAI

GPT-6 Luna

Budget

AI model capability comparison
SpecificationGPT-5.6 SolGPT-6 Luna
ProviderOpenAIOpenAI
TierFrontierBudget
Context window1.05M1.05M
Max output128K128K
Input / 1M tokens$4Winner: $0.10
Output / 1M tokens$20Winner: $0.50
WeightsClosedClosed
Parametersprevious-generation reasoning + coding modelGPT-6 low-cost tier
Reasoning levelsnone, low, medium, high, xhigh, maxnone, low, medium, high, xhigh, max
Modalitiestext, imagetext, image
API model idgpt-5.6-solgpt-6-luna
ReleasedJuly 9, 2026September 22, 2026
Artificial Analysis Intelligence Index [max] (2026-09-26)Winner: 4737.3
Terminal-Bench 2.1 (2026-07-09)88.8Not verifiedUnverified
GPQA Diamond (2026-07-09)94.6Not verifiedUnverified
DeepSWE 1.1 [max] (2026-09-22)Not verifiedUnverified66.6

Prices are USD per million tokens at standard rates, excluding batch and caching discounts. Bold indicates the better figure where one is objectively better. Values we could not confirm from the provider are shown as “Not verified” rather than estimated.

Benchmark receipts

Where each score comes from, and how far it can be compared across models.

  • 2026-09-26
    Artificial Analysis

    Composite index blending reasoning, knowledge, and coding evals into one 0–100 score.

    Comparable with caveatOnly same-version scores are comparable. v4.3.2 was rebaselined (the top score fell from 66 on v4.1.1 to ~58), so v4.1.1 figures must not be compared with v4.3.2 figures — check measuredAt.

  • 2026-09-22
    OpenAI — Introducing GPT-6 Sol and Luna (vendor; not yet on Datacurve's board)

    Agentic software engineering: pass@1 on real repo tasks with reported cost per completed task.

    Directly comparable

  • 2026-07-09
    OpenAI — GPT-5.6 evals table (vendor)

    Agentic terminal work: multi-step tasks executed in a sandboxed shell environment.

    Comparable with caveatNot comparable with Terminal-Bench 3.0 or 4.0 (different task sets) or v1; harness configuration (container, time limits) also shifts results.

Pricing tiers

GPT-5.6 Sol: Launch list was $5/$30; cut to $4/$20 on 2026-08-21, promo held at least through 2026-11-21. Above 272K input the whole request bills 2x input / 1.5x output ($8/$30). Cached input $0.40/MTok; cache writes 1.25x input; batch $2/$10. Knowledge cutoff Feb 16 2026.

GPT-6 Luna: Direct successor to GPT-5.6 Luna at half the price. Cached input $0.01/MTok; cache writes $0.125. Above 272K input the whole request bills $0.20/$0.75 ($0.02 cached). Batch/Flex $0.05/$0.25; Fast mode $0.20/$1.00. Knowledge cutoff May 18 2026.

FrontierRecord checked September 26, 2026

GPT-5.6 Sol

GPT-5.6 Sol was OpenAI's GPT-5.6 flagship for complex reasoning and coding with a 1.05M-token context — superseded on 2026-09-22 by GPT-6 Sol at half the price.

Best for

  • Complex reasoning
  • Agentic coding
  • Hard knowledge work

Watch out

Superseded: GPT-6 Sol lists $2/$10, half this row's $4/$20 promo rate, and scores higher on the Artificial Analysis index. Still served; 'max' effort and 'ultra' multi-agent mode multiply token use.

BudgetRecord checked September 26, 2026

GPT-6 Luna

GPT-6 Luna is OpenAI's cheapest current model at $0.10/$0.50 — the direct successor to GPT-5.6 Luna at half the price, with the full 1.05M-token context.

Best for

  • High-volume chat and extraction
  • Classification
  • Cost-capped agent loops

Watch out

Small tier: it matches GPT-5.6 Luna on the Artificial Analysis index but sits well below GPT-6 Sol on hard reasoning; the DeepSWE figure is vendor-reported at max effort.

When the cheaper one wins

GPT-6 Luna is cheaper on output at $0.50 per million tokens against $20 for GPT-5.6 Sol — about 40×. Use the cheaper tier for classification, extraction, summarisation, and any task where the expensive model’s extra score does not change the accepted output. The expensive one only pays if your hardest task actually fails on the cheap tier. These are standard-tier API rates, excluding batch and cache discounts.

Run the model picker

Evidence confidence: High

How strong and complete the evidence behind this comparison is — not a prediction of which model is better.

  • Pricing verified on both sides — Input and output rates are verified for both models.
  • 5/5 core specs verified on both sides — All core specifications verified for both models.
  • 1 shared named benchmark with differing scores — Measured on: Artificial Analysis Intelligence Index.
  • Verified within the last 90 days — Newest catalog check was 2 days ago.
  • Both models carry source citations — Each side has at least two catalog sources on record.

Source receipts

Each catalog figure was checked against the provider or an independent second source on the date shown.

Related comparisons

Diving deeper on one model? GPT-5.6 Sol · GPT-6 Luna

Common questions

GPT-5.6 Sol vs GPT-6 Luna

Answered from the verified figures on this page rather than general guidance.

Is GPT-5.6 Sol or GPT-6 Luna cheaper for input?
GPT-6 Luna is cheaper at $0.10 per million input tokens, against $4 for GPT-5.6 Sol — roughly 40× the price. Output tokens usually dominate a real bill, so weigh the output rate more heavily than the input rate. These are base rates; GPT-5.6 Sol has tiered pricing: Launch list was $5/$30; cut to $4/$20 on 2026-08-21, promo held at least through 2026-11-21. Above 272K input the whole request bills 2x input / 1.5x output ($8/$30). Cached input $0.40/MTok; cache writes 1.25x input; batch $2/$10. Knowledge cutoff Feb 16 2026. GPT-6 Luna has tiered pricing: Direct successor to GPT-5.6 Luna at half the price. Cached input $0.01/MTok; cache writes $0.125. Above 272K input the whole request bills $0.20/$0.75 ($0.02 cached). Batch/Flex $0.05/$0.25; Fast mode $0.20/$1.00. Knowledge cutoff May 18 2026.
Is GPT-5.6 Sol or GPT-6 Luna cheaper for output?
GPT-6 Luna is cheaper at $0.50 per million output tokens, against $20 for GPT-5.6 Sol — roughly 40× the price. Output tokens usually dominate a real bill, so weigh the output rate more heavily than the input rate. These are base rates; GPT-5.6 Sol has tiered pricing: Launch list was $5/$30; cut to $4/$20 on 2026-08-21, promo held at least through 2026-11-21. Above 272K input the whole request bills 2x input / 1.5x output ($8/$30). Cached input $0.40/MTok; cache writes 1.25x input; batch $2/$10. Knowledge cutoff Feb 16 2026. GPT-6 Luna has tiered pricing: Direct successor to GPT-5.6 Luna at half the price. Cached input $0.01/MTok; cache writes $0.125. Above 272K input the whole request bills $0.20/$0.75 ($0.02 cached). Batch/Flex $0.05/$0.25; Fast mode $0.20/$1.00. Knowledge cutoff May 18 2026.
Which has the larger context window, GPT-5.6 Sol or GPT-6 Luna?
Both accept about 1.05M tokens of context, so document length will not decide between them.
Do GPT-5.6 Sol and GPT-6 Luna support the same reasoning levels?
Yes — both accept the same effort settings: "none", "low", "medium", "high", "xhigh", "max". Higher effort costs more and takes longer, so start low and raise it only where output quality actually improves.
Should I use GPT-5.6 Sol or GPT-6 Luna?
GPT-5.6 Sol is the frontier tier and GPT-6 Luna the budget tier. The useful question is whether your hardest task actually fails on the cheaper one — most production volume such as classification, extraction and summarisation does not.

Next step

Choosing between them

The verified figures that separate this pair, computed from the catalog rather than restated boilerplate.

  • Input price: GPT-6 Luna costs $0.10 per 1M tokens versus $4 for GPT-5.6 Sol — a 40x difference at the headline tier.
  • Measured capability: GPT-5.6 Sol leads Artificial Analysis Intelligence Index 47 to 37.3 (measured 2026-09-26).
  • Positioning: GPT-5.6 Sol sits in the frontier tier, GPT-6 Luna in the budget tier — most production volume (classification, extraction, summarisation) does not need the pricier tier.
Run the model pickerEstimate stack costAll model comparisons

Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.