Inference
EN

GPT-6 Luna vs GPT-6 Sol

Change models
GPT-6 LunaOpenAI
GPT-6 SolOpenAI
IndexAll categories, out of 100.
65.5 ±2.1
72.8 ±2.2 best
GapIndex points below the top.
−7.3
Top
CoverageIndex weight that has results.
100%
100%
SpeedOutput tokens per second.
66/s best
61/s
EloLMArena rating and rank.
1388
1397 best
ReleasedMonth of release.
2026-09
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.10 best
$2.00
Output / 1MUS dollars per 1M output tokens.
$0.50 best
$10.00
Batch output / 1MBatch price per 1M output tokens.
$0.25 best
$5.00
1K chats2K tokens in, 500 out, each.
$0.45 best
$9.00
1K documents20K tokens in, 1K out, each.
$2.50 best
$50.00
1K agent tasks200K tokens in, 20K out, each.
$30.00 best
$600.00
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
66.9
68.6 best
CodingCode writing and repair.
66.8
73.9 best
ReasoningLogic problems and puzzles.
64.3
76.1 best
MultimodalTasks with images and text.
65.0
70.1 best
KnowledgeFacts and expert knowledge.
63.5
71.9 best
MultilingualTasks in many languages.
53.3
65.5 best
InstructionTasks with strict prompt rules.
39.8
65.3 best
MathMath problems.
68.5
75.8 best

Benchmarks

86 shared
Only GPT-6 Luna: 15Only GPT-6 Sol: 8
GPT-6 Luna
GPT-6 Sol
Artificial Analysis Humanity's Last ExamKnowledge
38.5%54 / 222
47.9% best18 / 222
Artificial Analysis Long Context ReasoningReasoning
83.3%13 / 217
83.7% best10 / 217
GPQA diamond · Epoch AIKnowledge
90.5%39 / 217
94.3% best10 / 217
Critical Physics TasksReasoning
19.4%31 / 214
30.9% best8 / 214
EuroEval PolishMultilingual
56.0%36 / 209
62.4% best14 / 209
OTIS Mock AIME 2024-2025Math
98.9%22 / 200
100.0% best1 / 200
EuroEval SwedishMultilingual
61.2%31 / 195
71.3% best6 / 195
EuroEval GermanMultilingual
50.8%36 / 191
63.1% best6 / 191
EuroEval PortugueseMultilingual
60.6%29 / 182
70.1% best6 / 182
EuroEval FrenchMultilingual
62.7%29 / 178
70.4% best9 / 178
EuroEval SpanishMultilingual
52.1%38 / 176
65.6% best7 / 176
EuroEval DutchMultilingual
59.0%29 / 175
67.5% best7 / 175
EuroEval ItalianMultilingual
62.1%24 / 174
69.9% best8 / 174
Artificial Analysis Omniscience AccuracyKnowledge
39.6%54 / 164
54.5% best13 / 164
GDPval-AA normalizedAgentic
46.6%36 / 133
50.4% best27 / 133
ARC-AGI-2 (semi-private)Reasoning
59.3%44 / 131
89.6% best10 / 131
ARC-AGI-1 (semi-private)Reasoning
86.7%46 / 126
95.5% best17 / 126
Vibe Code Bench v1.1Coding
81.6%26 / 123
87.8% best12 / 123
ARC-AGI-2 (public eval)Reasoning
61.8%36 / 122
90.0% best17 / 122
ARC-AGI-1 (public eval)Reasoning
92.5%40 / 116
97.8% best14 / 116