OpenAI: GPT-6.1 Sol

OpenAI

flagship

Capabilities

VisionToolsStructured OutputExtended ThinkingLong Context

Strengths

Exceptional reasoning and analysis capability (0.86)
Massive 1M+ token context for document-heavy workflows
Multimodal vision and file processing strengths
Extended thinking and structured output support
Agentic coding and computer use capabilities

Weaknesses

Extremely slow inference (0.03 speed score, ~51 tok/s)
High output pricing ($10/M) limits cost-sensitive applications
Closed-source with minimal transparency on training
Positioned below Astra in capability hierarchy
Substantial environmental footprint from large-scale inference

Pricing

Input / 1M tokens

Output / 1M tokens

$10

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 model with limited public methodology disclosure. OpenAI provides minimal training details; no access to weights, training data, or reproducibility information. Transparency aligned with OpenAI baseline.

Sustainability

Inference energy

N/A

Training footprint

N/A

Provider infrastructure

5.0 / 10

Composite:4.5 / 10

Large flagship model with significant inference compute requirements. OpenAI's sustainability commitments are not publicly detailed to the level of competitors like Google. No specific green infrastructure disclosures available.

Task Fitness

question answering

86%

code

87%

mathematics

50%

business communication

50%

vision

82%

analysis

86%

extraction

89%

long-form generation

85%

research

50%

reasoning

86%

summarisation

88%

translation

84%

conversation

85%