OpenAI: GPT-5.6 Sol

OpenAI

flagship

Capabilities

VisionToolsStructured OutputExtended ThinkingLong ContextCode

Strengths

Exceptional coding and reasoning performance
Multimodal vision capabilities integrated seamlessly
1.05M token context for long-document understanding
Extended thinking and agentic workflow support
Strong across nearly all task categories

Weaknesses

Extremely slow inference (0.07 speed score)
Very high output pricing ($30/M tokens)
Closed-source; no transparency or self-hosting
Opaque data governance and privacy practices
High compute cost per request limits cost-sensitive use

Pricing

Input / 1M tokens

Output / 1M tokens

$30

Context window

1,050k

Transparency

Open weights

0.0 / 10

Open training data

0.0 / 10

Open methodology

2.5 / 10

Licence openness

1.0 / 10

Provider disclosure

3.0 / 10

FMTI company score

N/A

Composite:3.5 / 10

Closed-source proprietary model. OpenAI publishes limited methodology details. No public weights, training data, or auditability. Transparency is constrained by commercial confidentiality.

Sustainability

Inference energy

N/A

Training footprint

N/A

Provider infrastructure

6.0 / 10

Composite:2.3 / 10

Large flagship model with significant inference compute overhead. OpenAI's sustainability commitments are unclear; no public carbon accounting for inference. Extremely long context window and extended thinking increase per-request energy cost.

Task Fitness

question answering

91%

code

93%

mathematics

87%

business communication

91%

vision

85%

analysis

92%

extraction

86%

long-form generation

91%

research

92%

reasoning

91%

summarisation

88%

translation

88%

conversation

91%