Compare · ModelsLive · 3 picked · head to head
GPT-5.6 Sol vs GPT-6.1 Sol vs GPT-6 Astra
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-6 Astra wins on 21/30 benchmarks
GPT-6 Astra wins 21 of 30 shared benchmarks. Leads in agentic · knowledge · general.
Category leads
agentic·GPT-6 Astrareasoning·GPT-6.1 Solknowledge·GPT-6 Astrageneral·GPT-6 Astracoding·GPT-6 Astramath·GPT-6.1 Solspeed·GPT-6 Astra
Hype vs Reality
Attention vs performance
GPT-5.6 Sol
#31 by perf·#10 by attention
GPT-6.1 Sol
#6 by perf·#5 by attention
GPT-6 Astra
#10 by perf·#5 by attention
Best value
GPT-6.1 Sol
1.2x better value than GPT-5.6 Sol
GPT-5.6 Sol
11.2 pts/$
$6.00/M
GPT-6.1 Sol
12.9 pts/$
$6.00/M
GPT-6 Astra
2.5 pts/$
$30.00/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
30 benchmarks · 3 models
GPT-5.6 SolGPT-6.1 SolGPT-6 Astra
APEX-Agents
GPT-6 Astra leads by +4.7
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
GPT-5.6 Sol
51.4
GPT-6.1 Sol
60.0
GPT-6 Astra
64.7
ARC-AGI
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
GPT-5.6 Sol
97.5
GPT-6.1 Sol
98.5
GPT-6 Astra
98.5
ARC-AGI-2
GPT-6 Astra leads by +0.8
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
GPT-5.6 Sol
92.5
GPT-6.1 Sol
94.2
GPT-6 Astra
95.0
Chess Puzzles
GPT-6 Astra leads by +6.3
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5.6 Sol
66.3
GPT-6.1 Sol
66.3
GPT-6 Astra
72.6
Ebr Bench
GPT-6 Astra leads by +21.9
GPT-5.6 Sol
44.8
GPT-6.1 Sol
54.3
GPT-6 Astra
76.2
Frontiercode
GPT-6 Astra leads by +3.0
GPT-5.6 Sol
47.5
GPT-6.1 Sol
50.2
GPT-6 Astra
53.3
FrontierMath-Tier-4-v2-Private
GPT-6.1 Sol leads by +2.4
GPT-5.6 Sol
82.9
GPT-6.1 Sol
100.0
GPT-6 Astra
97.6
FrontierMath-Tiers-1-3-v2-Private
GPT-5.6 Sol
89.1
GPT-6.1 Sol
93.7
GPT-6 Astra
93.7
Furniture Assembly
GPT-5.6 Sol
38.1
GPT-6.1 Sol
71.4
GPT-6 Astra
71.4
GPQA diamond
GPT-6 Astra leads by +0.5
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5.6 Sol
91.3
GPT-6.1 Sol
93.9
GPT-6 Astra
94.4
Mystery Game Puzzles
GPT-6 Astra leads by +4.4
GPT-5.6 Sol
53.7
GPT-6.1 Sol
78.0
GPT-6 Astra
82.4
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5.6 Sol
100.0
GPT-6.1 Sol
100.0
GPT-6 Astra
100.0
Proofbench
GPT-5.6 Sol
83.0
GPT-6.1 Sol
99.0
GPT-6 Astra
99.0
SimpleQA Verified
GPT-6 Astra leads by +1.7
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
GPT-5.6 Sol
69.7
GPT-6.1 Sol
73.9
GPT-6 Astra
75.6
Artificial Analysis · CritPt
GPT-6.1 Sol
31.7
GPT-6 Astra
31.7
Artificial Analysis · GDPval
GPT-6.1 Sol leads by +1.1
GPT-6.1 Sol
54.7
GPT-6 Astra
53.6
Artificial Analysis · Humanity's Last Exam
GPT-6 Astra leads by +1.8
GPT-6.1 Sol
52.9
GPT-6 Astra
54.7
Artificial Analysis · Long Context Reasoning
GPT-6.1 Sol leads by +2.3
GPT-6.1 Sol
83.0
GPT-6 Astra
80.7
Artificial Analysis · MMMU Pro
GPT-6 Astra leads by +0.9
GPT-6.1 Sol
86.0
GPT-6 Astra
86.9
Artificial Analysis · Quality Index
GPT-6 Astra leads by +0.8
GPT-6.1 Sol
51.8
GPT-6 Astra
52.7
Artificial Analysis · SciCode
GPT-6 Astra leads by +2.3
GPT-6.1 Sol
54.2
GPT-6 Astra
56.5
Balrog
GPT-6 Astra leads by +8.3
Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning.
GPT-5.6 Sol
60.0
GPT-6 Astra
68.3
Deepswe
GPT-6 Astra leads by +1.5
GPT-5.6 Sol
72.7
GPT-6 Astra
74.1
Dtbench
GPT-6 Astra leads by +2.2
GPT-5.6 Sol
93.3
GPT-6 Astra
95.5
Frontierswe
GPT-6 Astra leads by +33.3
GPT-5.6 Sol
32.2
GPT-6 Astra
65.5
GSO-Bench
GPT-6 Astra leads by +2.9
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
GPT-5.6 Sol
76.5
GPT-6 Astra
79.4
Lmca
GPT-6 Astra leads by +6.1
GPT-5.6 Sol
69.7
GPT-6 Astra
75.7
Mirrorcode
GPT-6 Astra leads by +26.7
GPT-5.6 Sol
20.0
GPT-6 Astra
46.7
SpireClimb
GPT-6 Astra leads by +1.0
GPT-6.1 Sol
24.7
GPT-6 Astra
25.7
WeirdML
GPT-6 Astra leads by +4.1
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5.6 Sol
89.4
GPT-6 Astra
93.6
Full benchmark table
| Benchmark | GPT-5.6 Sol | GPT-6.1 Sol | GPT-6 Astra |
|---|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 51.4 | 60.0 | 64.7 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 97.5 | 98.5 | 98.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 92.5 | 94.2 | 95.0 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 66.3 | 66.3 | 72.6 |
Ebr Bench | 44.8 | 54.3 | 76.2 |
Frontiercode | 47.5 | 50.2 | 53.3 |
FrontierMath-Tier-4-v2-Private | 82.9 | 100.0 | 97.6 |
FrontierMath-Tiers-1-3-v2-Private | 89.1 | 93.7 | 93.7 |
Furniture Assembly | 38.1 | 71.4 | 71.4 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 91.3 | 93.9 | 94.4 |
Mystery Game Puzzles | 53.7 | 78.0 | 82.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 100.0 | 100.0 | 100.0 |
Proofbench | 83.0 | 99.0 | 99.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 69.7 | 73.9 | 75.6 |
Artificial Analysis · CritPt | — | 31.7 | 31.7 |
Artificial Analysis · GDPval | — | 54.7 | 53.6 |
Artificial Analysis · Humanity's Last Exam | — | 52.9 | 54.7 |
Artificial Analysis · Long Context Reasoning | — | 83.0 | 80.7 |
Artificial Analysis · MMMU Pro | — | 86.0 | 86.9 |
Artificial Analysis · Quality Index | — | 51.8 | 52.7 |
Artificial Analysis · SciCode | — | 54.2 | 56.5 |
Balrog Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning. | 60.0 | — | 68.3 |
Deepswe | 72.7 | — | 74.1 |
Dtbench | 93.3 | — | 95.5 |
Frontierswe | 32.2 | — | 65.5 |
GSO-Bench GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues. | 76.5 | — | 79.4 |
Lmca | 69.7 | — | 75.7 |
Mirrorcode | 20.0 | — | 46.7 |
SpireClimb | — | 24.7 | 25.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 89.4 | — | 93.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $2.00 | $10.00 | 1.1M tokens (~525 books) | $40.00 | |
| $2.00 | $10.00 | 1.1M tokens (~525 books) | $40.00 | |
| $10.00 | $50.00 | 1.1M tokens (~525 books) | $200.00 |