Compare · ModelsLive · 2 picked · head to head
GPT-5.5 vs GPT-5.6 Sol
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5.6 Sol wins on 19/24 benchmarks
GPT-5.6 Sol wins 19 of 24 shared benchmarks. Leads in reasoning · knowledge · coding.
Category leads
agentic·GPT-5.5reasoning·GPT-5.6 Solknowledge·GPT-5.6 Solcoding·GPT-5.6 Solgeneral·GPT-5.6 Solmath·GPT-5.6 Sol
Head to head · from the data · data as of 2026-11-17
- Benchmarks · GPT-5.5 leads on 3 of 24 shared benchmarks (APEX-Agents, GPQA diamond, Mirrorcode), GPT-5.6 Sol on 19 (ARC-AGI, ARC-AGI-2, Chess Puzzles, Deepswe, 15 more) · 2 tied
- Price · GPT-5.5 input $5.00/M vs GPT-5.6 Sol $4.00/M, GPT-5.6 Sol is 20% cheaper (OpenRouter list price, as of 2026-11-17)
- Cheapest provider · GPT-5.5 at OpenAI $2.50/M input (of 7 offers) as of 2026-11-17 · GPT-5.6 Sol at OpenAI $2.00/M input (of 8 offers) as of 2026-11-17
- Context · GPT-5.5 and GPT-5.6 Sol both 1.1M tokens
- Released · GPT-5.5 Apr 24, 2026 vs GPT-5.6 Sol Jul 9, 2026 · 3 months apart
- Price history · GPT-5.5 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-17
Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.
Hype vs Reality
Attention vs performance
GPT-5.5
#64 by perf·#11 by attention
GPT-5.6 Sol
#36 by perf·#11 by attention
Best value
GPT-5.6 Sol
1.7x better value than GPT-5.5
GPT-5.5
3.3 pts/$
$17.50/M
GPT-5.6 Sol
5.5 pts/$
$12.00/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
24 benchmarks · 2 models
GPT-5.5GPT-5.6 Sol
APEX-Agents
GPT-5.5 leads by +3.7
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
GPT-5.5
55.1
GPT-5.6 Sol
51.4
ARC-AGI
GPT-5.6 Sol leads by +2.5
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
GPT-5.5
95.0
GPT-5.6 Sol
97.5
ARC-AGI-2
GPT-5.6 Sol leads by +7.5
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
GPT-5.5
85.0
GPT-5.6 Sol
92.5
Chess Puzzles
GPT-5.6 Sol leads by +5.3
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5.5
61.1
GPT-5.6 Sol
66.3
Deepswe
GPT-5.6 Sol leads by +5.6
GPT-5.5
67.0
GPT-5.6 Sol
72.7
Dtbench
GPT-5.5
93.3
GPT-5.6 Sol
93.3
Ebr Bench
GPT-5.6 Sol leads by +10.5
GPT-5.5
34.3
GPT-5.6 Sol
44.8
Frontiercode
GPT-5.6 Sol leads by +4.5
GPT-5.5
43.0
GPT-5.6 Sol
47.5
FrontierMath-Tier-4-v2-Private
GPT-5.6 Sol leads by +10.4
GPT-5.5
72.5
GPT-5.6 Sol
82.9
FrontierMath-Tiers-1-3-v2-Private
GPT-5.6 Sol leads by +4.1
GPT-5.5
89.8
GPT-5.6 Sol
93.8
Furniture Assembly
GPT-5.6 Sol leads by +17.9
GPT-5.5
20.2
GPT-5.6 Sol
38.1
GPQA diamond
GPT-5.5 leads by +0.7
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5.5
95.8
GPT-5.6 Sol
95.1
GSO-Bench
GPT-5.6 Sol leads by +36.3
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
GPT-5.5
40.2
GPT-5.6 Sol
76.5
Lmca
GPT-5.6 Sol leads by +5.8
GPT-5.5
63.9
GPT-5.6 Sol
69.7
Mirrorcode
GPT-5.5 leads by +3.3
GPT-5.5
23.3
GPT-5.6 Sol
20.0
Mystery Game Puzzles
GPT-5.6 Sol leads by +2.2
GPT-5.5
51.5
GPT-5.6 Sol
53.7
Osworld 2 0
GPT-5.6 Sol leads by +14.3
GPT-5.5
13.0
GPT-5.6 Sol
27.3
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5.5
100.0
GPT-5.6 Sol
100.0
PostTrainBench
GPT-5.6 Sol leads by +7.1
GPT-5.5
23.9
GPT-5.6 Sol
31.0
Proofbench
GPT-5.6 Sol leads by +33.0
GPT-5.5
50.0
GPT-5.6 Sol
83.0
SimpleBench
GPT-5.6 Sol leads by +3.2
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
GPT-5.5
62.8
GPT-5.6 Sol
66.0
SimpleQA Verified
GPT-5.6 Sol leads by +6.7
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
GPT-5.5
63.0
GPT-5.6 Sol
69.7
Surface Evolver Bench
GPT-5.6 Sol leads by +5.0
GPT-5.5
88.1
GPT-5.6 Sol
93.1
WeirdML
GPT-5.6 Sol leads by +4.5
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5.5
84.9
GPT-5.6 Sol
89.4
Full benchmark table
| Benchmark | GPT-5.5 | GPT-5.6 Sol |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 55.1 | 51.4 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 95.0 | 97.5 |
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. | 85.0 | 92.5 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 61.1 | 66.3 |
Deepswe | 67.0 | 72.7 |
Dtbench | 93.3 | 93.3 |
Ebr Bench | 34.3 | 44.8 |
Frontiercode | 43.0 | 47.5 |
FrontierMath-Tier-4-v2-Private | 72.5 | 82.9 |
FrontierMath-Tiers-1-3-v2-Private | 89.8 | 93.8 |
Furniture Assembly | 20.2 | 38.1 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 95.8 | 95.1 |
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. | 40.2 | 76.5 |
Lmca | 63.9 | 69.7 |
Mirrorcode | 23.3 | 20.0 |
Mystery Game Puzzles | 51.5 | 53.7 |
Osworld 2 0 | 13.0 | 27.3 |
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 |
PostTrainBench | 23.9 | 31.0 |
Proofbench | 50.0 | 83.0 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 62.8 | 66.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 63.0 | 69.7 |
Surface Evolver Bench | 88.1 | 93.1 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 84.9 | 89.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $5.00 | $30.00 | 1.1M tokens (~525 books) | $112.50 | |
| $4.00 | $20.00 | 1.1M tokens (~525 books) | $80.00 |
GPT-5.5 vs GPT-5.6 Sol · questions answered from the data
GPT-5.5 leads on 3 of 24 shared benchmarks (APEX-Agents, GPQA diamond, Mirrorcode), GPT-5.6 Sol on 19 (ARC-AGI, ARC-AGI-2, Chess Puzzles, Deepswe, 15 more) · 2 tied