GPT-5.6 Sol AI Evaluation | LayerLens - Benchmark Results & Performance

Model

GPT-5.6 Sol by OpenAI

text+image+file->text

Free credits included. No credit card required.

GPT-5.6 Sol is OpenAI's flagship model in the GPT-5.6 series, designed for complex reasoning, coding, and agentic workflows. It is particularly strong at command-line and multi-step coding tasks, long-horizon cybersecurity work including vulnerability research, and advanced biology workflows. The model introduces a new max reasoning effort mode and an ultra mode that leverages subagents to accelerate complex work. GPT-5.6 Sol launched as a limited preview to trusted partners, with broader availability planned in the coming weeks.

Evaluations

across 10 benchmarks

Median latency

4.2s

Context length

tokens

Cost

$5 · $30

input · output per 1M tokens

Key takeaways

Key takeaways will appear here once evaluations are run.

Technical specifications

Context length: 1M tokens
Max output: —
Architecture: MoE
Modalities: text+image+file->text
Parameters: —
Open weights: No
Licence: Proprietary
Regions: USA
Provider: OpenAI
Provider model key: openai/gpt-5.6-sol