modelbench日本語

GPT-6 Luna

OpenAICurrent

Input price
$0.10
Per 1 million tokens (USD)
Output price
$0.50
Per 1 million tokens (USD)
Effort levels
6total
Default: Med
Benchmark results
26total
Benchmark types: 18

Model information

API model ID
gpt-6-luna
release date
2026-09-22
effort
None / Low / Med (default) / High / Xhigh / Max
Effort source
Based on reasoning effort on the OpenAI model page (developers.openai.com/api/docs/models/gpt-6-luna).
Claude Code
unavailable
Codex
available
Availability notes
Codex desktop app / CLI / IDE. 'GPT-6 Luna supports reasoning efforts up to Max, but not Ultra.' Enterprise/Edu requires administrator enablement. Claude Code's official model list (code.claude.com/docs/en/model-config alias table) contains only Anthropic models, so claudeCode is false (connections via a gateway are not counted as official options).
Availability source
https://learn.chatgpt.com/docs/models

Price (USD per 1 million tokens)

Prices do not vary by effort; token consumption does.

input$0.10
output$0.50
cache read$0.01
cache write$0.13

Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request. Batch/Flex 50%, Fast 2x.

| gpt-6-luna | $0.10 | $0.50 | $0.01 | $0.125 | $0.25 | $0.20 | $0.75 | $0.02 | $0.25 | $0.50 |

Source: https://developers.openai.com/api/docs/pricing

Benchmark results

BenchmarkeffortScoreSource cost (metric)measured/reported bysourceConditions and extraction method
Terminal-Bench 4.0not statednot stated16.4%—not statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. table note: see the Haiku 5.5 System Card (effort not stated); values for other vendor models (as listed by Anthropic)transcribed from table
Source excerpt16.4%Location Table, Terminal-Bench 4.0 row: 16.4
FrontierCode v1.1(main set)not statednot stated42.4%—not statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. table note: see the Haiku 5.5 System Card (effort not stated); values for other vendor models (as listed by Anthropic)transcribed from table
Source excerpt42.4%Location Table, FrontierCode v1.1 (Main) row: 42.4
GDPval-AA v2.1LowLow1,036$0.004Cost per tasknot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "GDPval-AA v2.1 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Low" was mapped to low in the OpenAI API reasoning efforttranscribed from text
Source excerptGDPval-AA v2.1 — Accuracy vs. cost: "GPT-6 Luna · Low: 1036.0 $0.004"
GDPval-AA v2.1Med (default)Med1,262$0.02Cost per tasknot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "GDPval-AA v2.1 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Med" was mapped to medium in the OpenAI API reasoning efforttranscribed from text
Source excerptGDPval-AA v2.1 — Accuracy vs. cost: "GPT-6 Luna · Med: 1262.0 $0.02"
GDPval-AA v2.1HighHigh1,344$0.03Cost per tasknot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "GDPval-AA v2.1 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "High" was mapped to high in the OpenAI API reasoning efforttranscribed from text
Source excerptGDPval-AA v2.1 — Accuracy vs. cost: "GPT-6 Luna · High: 1344.0 $0.03"
GDPval-AA v2.1XhighXhigh1,364$0.05Cost per tasknot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "GDPval-AA v2.1 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Xhigh" was mapped to xhigh in the OpenAI API reasoning efforttranscribed from text
Source excerptGDPval-AA v2.1 — Accuracy vs. cost: "GPT-6 Luna · Xhigh: 1364.0 $0.05"
GDPval-AA v2.1MaxMax1,437$0.09Cost per tasknot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "GDPval-AA v2.1 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Max" was mapped to max in the OpenAI API reasoning efforttranscribed from text
Source excerptGDPval-AA v2.1 — Accuracy vs. cost: "GPT-6 Luna · Max: 1437.0 $0.09"
AA-Briefcase v1.1(Elo)not statednot stated1,336—not statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. table note: see the Haiku 5.5 System Card (effort not stated); values for other vendor models (as listed by Anthropic)transcribed from table
Source excerpt1336Location Table, AA-Briefcase v1.1 (Elo) row: 1336
OSWorld 2.1(offline subset)LowLow19.2%$0.04Cost per attemptnot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "OSWorld 2.1 (offline subset) / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Low" was mapped to low in the OpenAI API reasoning efforttranscribed from text
Source excerptOSWorld 2.1 (offline subset) — Accuracy vs. cost: "GPT-6 Luna · Low: 19.2% $0.04"
OSWorld 2.1(offline subset)Med (default)Med37.5%$0.13Cost per attemptnot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "OSWorld 2.1 (offline subset) / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Med" was mapped to medium in the OpenAI API reasoning efforttranscribed from text
Source excerptOSWorld 2.1 (offline subset) — Accuracy vs. cost: "GPT-6 Luna · Med: 37.5% $0.13"
OSWorld 2.1(offline subset)HighHigh42.3%$0.14Cost per attemptnot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "OSWorld 2.1 (offline subset) / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "High" was mapped to high in the OpenAI API reasoning efforttranscribed from text
Source excerptOSWorld 2.1 (offline subset) — Accuracy vs. cost: "GPT-6 Luna · High: 42.3% $0.14"
OSWorld 2.1(offline subset)XhighXhigh44.8%$0.17Cost per attemptnot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "OSWorld 2.1 (offline subset) / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Xhigh" was mapped to xhigh in the OpenAI API reasoning efforttranscribed from text
Source excerptOSWorld 2.1 (offline subset) — Accuracy vs. cost: "GPT-6 Luna · Xhigh: 44.8% $0.17"
OSWorld 2.1(offline subset)MaxMax48.9%$0.20Cost per attemptnot statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. A point in the figure "OSWorld 2.1 (offline subset) / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Max" was mapped to max in the OpenAI API reasoning efforttranscribed from text
Source excerptOSWorld 2.1 (offline subset) — Accuracy vs. cost: "GPT-6 Luna · Max: 48.9% $0.20"
Chartography (without tools)not statednot stated29.1%—not statedIntroducing Claude Haiku 5.5official · Retrieved 2026-11-12Vendor-reported value. table note: see the Haiku 5.5 System Card (effort not stated); values for other vendor models (as listed by Anthropic)transcribed from table
Source excerpt29.1%Location Table, Chartography (no tools) row: 29.1
HealthBench Professional (length-adjusted score)not statednot stated60.8—OpenAIAddendum to GPT-6 Astra System Card: GPT-6.1 Terraofficial · Retrieved 2026-11-14Vendor-reported value. Self-reported by OpenAI. effort not stated. length-adjusted score. Parentheses contain the unadjusted score and mean response length in characters; only the leading value is used as score. Cost not stated.transcribed from table
Source excerpt60.8 (61.2, 2119)Location §5.1 Table 5, HealthBench Professional length-adjusted row, gpt-6-luna column
HealthBench Hard (length-adjusted score)not statednot stated31.4—OpenAIAddendum to GPT-6 Astra System Card: GPT-6.1 Terraofficial · Retrieved 2026-11-14Vendor-reported value. Self-reported by OpenAI. effort not stated. length-adjusted score. Parentheses contain the unadjusted score and mean response length in characters; only the leading value is used as score. Cost not stated.transcribed from table
Source excerpt31.4 (25.4, 1241)Location §5.1 Table 5, HealthBench Hard length-adjusted row, gpt-6-luna column
GPQA DiamondMaxMax90.5%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, gpqa_diamond.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptgpqa_diamond.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.9046717171717171, stderr=0.018537378631915503, Started at=2026-09-22T22:14:05.000Z
FrontierMath Tiers 1-3(v2)MaxMax79.0%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, frontiermath_tiers_1_3_v2.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptfrontiermath_tiers_1_3_v2.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.7894736842105263, stderr=0.024191503146153773, Started at=2026-09-22T22:14:05.000Z
FrontierMath Tier 4(v2)MaxMax56.1%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, frontiermath_tier_4_v2.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptfrontiermath_tier_4_v2.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.5609756097560976, stderr=0.07846686801046543, Started at=2026-09-22T22:14:05.000Z
OTIS Mock AIME 2024-2025MaxMax98.9%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, otis_mock_aime_2024_2025.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptotis_mock_aime_2024_2025.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.9888888888888889, stderr=0.006674027922425511, Started at=2026-09-22T22:14:05.000Z
SimpleQA VerifiedMaxMax41.4%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, simpleqa_verified.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptsimpleqa_verified.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.414, stderr=0.015583544104177553, Started at=2026-09-22T22:14:05.000Z
Chess Puzzles(Epoch AI)MaxMax36.0%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, chess_puzzles.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptchess_puzzles.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.36, stderr=0.04824181513244218, Started at=2026-10-13T13:37:06.000Z
MirrorCodeMaxMax26.7%—Epoch AIEpoch AI, Capabilities & benchmarking (benchmark_data.zip, mirrorcode.csv) CC BY 4.0third-party · Retrieved 2026-11-12Run by Epoch AI (third-party measurement). Source label "GPT-6 Luna (max)". effort is indicated by the suffix _max in Epoch's Model version. Data retrieved 2026-11-12.transcribed from table
Source excerptmirrorcode.csv: Model version=gpt-6-luna_max, Best score (across scorers)=0.26666666666666666, stderr=0.09026709338484401, Started at=2026-10-21T16:48:12.000Z
Epoch Capabilities Index(ECI)not statednot stated156.41—Epoch AIEpoch AI, epoch_capabilities_index/eci_scores.csv (CC BY 4.0)third-party · Retrieved 2026-11-12A per-model metric that combines results across multiple benchmarks and multiple efforts (not per effort). Calculated by Epoch AI.transcribed from table
Source excerpteci_scores.csv: Model=GPT-6 Luna, eci=156.41, eci_ci_low=154.45, eci_ci_high=158.5
LMArena Text (style control, overall)MaxMax1,441—LMArenalmarena-ai/leaderboard-dataset text_style_control/latest-00000-of-00001.parquet (CC BY 4.0)third-party · Retrieved 2026-11-12Human head-to-head votes (Bradley-Terry). Source label "GPT-6 Luna (Max)" (effort is in parentheses in the label). Release date 2026-11-09.transcribed from table
Source excerptmodel_name=GPT-6 Luna (Max), rating=1440.82, rating_lower=1436.7, rating_upper=1444.9, vote_count=107021, rank=98, category=overall, leaderboard_publish_date=2026-11-09
LMArena Text Coding (without style control)MaxMax1,432—LMArenalmarena-ai/leaderboard-dataset text/latest-00000-of-00001.parquet (CC BY 4.0)third-party · Retrieved 2026-11-12Human head-to-head votes (Bradley-Terry). Source label "GPT-6 Luna (Max)" (effort is in parentheses in the label). Release date 2026-11-09.transcribed from table
Source excerptmodel_name=GPT-6 Luna (Max), rating=1432.04, rating_lower=1426.3, rating_upper=1437.8, vote_count=24606, rank=139, category=coding, leaderboard_publish_date=2026-11-09

Related rankings

References