GoTry AI

GPT-5.6 Luna

Language OpenAI Available Released Jul 9, 2026 Partly verified
Model ID
Context
1.05M
128K max output
Input price
$0.20
per 1M tokens · OpenAI API
Output price
$1.20
per 1M tokens
Epoch Capabilities Index
156.3
#21 of 71 listed
LMArena Text
1,454
#41 of 79 listed

Where to run it

3 offerings · prices as listed
MeterTierPrice
OpenAI API (opens in a new tab) Official Global · USD gpt-5.6-luna
InputStandard $0.20 / 1M tokens
Batch $0.10 / 1M tokens
Flex tier $0.10 / 1M tokens
Fast tier $0.40 / 1M tokens
OutputStandard $1.20 / 1M tokens
Batch $0.60 / 1M tokens
Flex tier $0.60 / 1M tokens
Fast tier $2.40 / 1M tokens
Cache readStandard $0.020 / 1M tokens
Batch $0.010 / 1M tokens
Flex tier $0.010 / 1M tokens
Fast tier $0.040 / 1M tokens
Cache writeStandard $0.25 / 1M tokens
Batch $0.13 / 1M tokens
Flex tier $0.13 / 1M tokens
Fast tier $0.50 / 1M tokens
InputStandard $0.20 / 1M tokens
OutputStandard $1.20 / 1M tokens
Cache readStandard $0.020 / 1M tokens
Cache writeStandard $0.25 / 1M tokens
Azure OpenAI in Foundry (opens in a new tab) data-zone · USD gpt-5-6-luna
InputStandard $0.22 / 1M tokens
OutputStandard $1.32 / 1M tokens
Cache readStandard $0.030 / 1M tokens
Cache writeStandard $0.28 / 1M tokens
  1. 4 prices dropped on OpenAI API Input (Prompts over 272K tokens, Fast tier) $0.80 · Cache read (Prompts over 272K tokens, Fast tier) $0.080 · Cache write (Prompts over 272K tokens, Fast tier) $1.00 · Output (Prompts over 272K tokens, Fast tier) $3.60

Vendor-reported scores

Published by OpenAI, under the conditions it states
BenchmarkSettingScore
SWE-bench Pro—62.7%
DeepSWE 1.1—67.2%
Terminal-Bench 2.1single agent84.7%
OSWorld 2.0set and scoring not stated45.6%
BrowseCompsingle agent83.3%
GPQA Diamond—92.3%
MMMU-Prono tools78.4%
with tools79.5%
MRCR v28-needle, 256K–512K41.3%
8-needle, 512K–1M41.3%

Cited scores

From third-party leaderboards, not measured by us
  • Epoch Capabilities Index 156.3

    #21 of 71 listed models · CI 153.6–158.9

    Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. Licensed CC BY 4.0. · License · as of Sep 26, 2026 · changes we made

  • LMArena Text 1,454

    #41 of 79 listed models · xhigh effort · CI 1,450–1,459 · 35,823 votes

    LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Sep 25, 2026 · changes we made

  • Epoch AI · FrontierMath Tiers 1–3 82.1%

    #10 of 56 listed models · max effort · ±2.27

    Also listed: low effort 41.4% · reasoning off 39.6%

    Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. Licensed CC BY 4.0. · License · as of Sep 26, 2026 · changes we made

  • Epoch AI · GPQA Diamond 91.6%

    #18 of 64 listed models · max effort · ±1.73

    Also listed: low effort 82.3% · reasoning off 63.6%

    Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. Licensed CC BY 4.0. · License · as of Sep 26, 2026 · changes we made

  • LMArena Text · Chinese 1,476

    #52 of 79 listed models · xhigh effort · CI 1,464–1,488 · 2,848 votes

    LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Sep 25, 2026 · changes we made

  • LMArena Text · Coding 1,501

    #41 of 79 listed models · xhigh effort · CI 1,494–1,508 · 9,953 votes

    LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Sep 25, 2026 · changes we made

  • LMArena Vision 1,259

    #32 of 49 listed models · xhigh effort · CI 1,250–1,267 · 7,793 votes

    LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Sep 13, 2026 · changes we made