Model Hub
Comparable AI models with a version, source, and review date.
This directory distinguishes selectable API versions from editorial profiles. Explore each model's focus, official sources, and review date.
Model directory
OpenAI
GPT-4o
StableSelectable API model in Debatidor
An OpenAI API model for comparing answers, reviewing proposals, and joining a room with the user's BYOK key.
Editorial review:
View model profileAnthropic
Claude Sonnet 4.6
StableSelectable API model in Debatidor
An Anthropic API model for examining requirements, challenging decisions, and contributing a technical second opinion with BYOK.
Editorial review:
View model profileGemini 3.1 Flash-Lite
PreviewSelectable API model in Debatidor
A Google API preview model for comparing syntheses and proposals with BYOK; availability and limits depend on the user's account.
Editorial review:
View model profileDeepSeek
DeepSeek V4.1 Flash
StableSelectable API model in Debatidor
A DeepSeek API model offered as deepseek-flash for evaluating code solutions and technical arguments with BYOK.
Editorial review:
View model profileOpenAI
GPT-5.1
StableEditorial profile · no selectable API ID
A general-purpose model aimed at complex tasks, software development, and tool use. This is an editorial profile; its exact ID is not in Debatidor's room API catalog.
Editorial review:
View model profileAnthropic
Claude Opus 4.1
RetiredEditorial profile · no selectable API ID
A historical Anthropic model for complex reasoning and coding. Its API version was retired on August 5, 2026; this page remains an editorial reference without an inference CTA.
Editorial review:
View model profileGemini 3.1 Pro
PreviewEditorial profile · no selectable API ID
A Google model aimed at complex problems, multimodal understanding, and agentic workflows. Its preview status should be considered when evaluating production stability and availability.
Editorial review:
View model profileDeepSeek
DeepSeek V4 Pro
PreviewEditorial profile · no selectable API ID
A DeepSeek model with reasoning and non-reasoning modes, aimed at code and agent tasks. This editorial profile brings together documented features and criteria for evaluating it on your task.
Editorial review:
View model profileCompare two models
Choose two versions to compare their editorial profiles by the same criteria. Sources and review dates belong to each model; this is not an Arena Score or a test of availability in your account.
OpenAI · Stable
GPT-4o
Text, vision, and answer comparison
- Debatidor API ID
- gpt-4o
- Strengths to compare
- Text and code analysis · Multimodal provider input · Fast proposal comparison
- Cases to evaluate
- Quality of objections · Instruction following · Clarity of technical solutions
- Editorial review
Anthropic · Stable
Claude Sonnet 4.6
Analysis, coding, and critical review
- Debatidor API ID
- claude-sonnet-4-6
- Strengths to compare
- Requirements analysis · Code and plan review · Technical argumentation
- Cases to evaluate
- Weak assumptions · Quality of rebuttals · Consistency of proposals
- Editorial review
Available comparisons
Editorial profiles and a sourced ranking
These profiles bring together documented features and criteria for evaluating models on your task. Arena.ai's published ranks and ratings are in the ranking, with attribution, date, votes, and uncertainty.
The external leaderboard and editorial comparator answer different questions. Check the exact version and sources before choosing; prices, limits, and availability depend on the provider and your account.