Section H.1: Proprietary Model Families

The following models are available exclusively through commercial APIs. They represent the current frontier in terms of general capability, though the gap with open models continues to narrow.

OpenAI GPT-4o / GPT-4o mini

Proprietary Multimodal API Only

Parameters	Undisclosed (rumored ~200B for GPT-4o, ~8B for mini)
Context Length	128K tokens
Modalities	Text, image, audio input; text, audio output
Key Strengths	Strong general reasoning, coding, multimodal understanding, fast inference (4o mini)
API Pricing	GPT-4o: $2.50/$10.00 per 1M input/output tokens; mini: $0.15/$0.60
Best For	General-purpose applications, multimodal tasks, production systems needing reliability

OpenAI o1 / o3 / o4-mini (Reasoning Models)

Proprietary Reasoning API Only

Parameters	Undisclosed
Context Length	200K tokens (o3, o4-mini); 128K (o1)
Architecture	Extended chain-of-thought reasoning with hidden "thinking" tokens
Key Strengths	Complex math, science, coding competitions, multi-step logical reasoning
API Pricing	o3: $10/$40 per 1M tokens; o4-mini: $1.10/$4.40 per 1M tokens
Best For	Hard reasoning tasks (math olympiads, research-level science, complex code generation)

Anthropic Claude 3.5 Sonnet / Claude 4 (Opus, Sonnet)

Proprietary Multimodal API + Web

Parameters	Undisclosed
Context Length	200K tokens
Modalities	Text, image, PDF input; text output (with tool use and computer use)
Key Strengths	Long-context reasoning, careful instruction following, coding, agentic tool use, safety
API Pricing	Claude 4 Sonnet: $3/$15 per 1M tokens; Claude 4 Opus: $15/$75 per 1M tokens
Best For	Long document analysis, coding agents, applications requiring nuanced safety, extended conversations

Google Gemini 2.0 Flash / Gemini 2.5 Pro

Proprietary Multimodal API + Web

Parameters	Undisclosed (Gemini 2.0 Flash is a smaller, faster variant)
Context Length	1M tokens (2.5 Pro); 1M tokens (2.0 Flash)
Modalities	Text, image, video, audio input; text, image output
Key Strengths	Extremely long context, native multimodal processing, competitive reasoning (2.5 Pro), speed (Flash)
API Pricing	2.0 Flash: $0.10/$0.40 per 1M tokens; 2.5 Pro: $1.25/$10.00 per 1M tokens
Best For	Very long documents, video/audio analysis, cost-sensitive multimodal applications

Pricing and Availability

API pricing and rate limits change frequently. Always check the provider's current pricing page before committing to a model for production use.