OpenAI (GPT)

The Majestix AI Inference Hub provides access to four GPT chat models and one image generation model from OpenAI. The GPT-5 family offers 400K token context windows -- double that of Claude models -- making them well-suited for processing large documents alongside conversation history.

Available Models

Key	Underlying Model	Context Window	Category
gpt-5-mini	GPT-5 Mini	400K	Fast / General
gpt-5.2	GPT-5.2	400K	Flagship
gpt-5.2-codex	GPT-5.2 Codex	400K	Coding / Agentic
cmo-agent	CMO Agent (fine-tuned)	128K	Marketing Specialist
gpt-5-image	GPT-5 Image	--	Image Generation

gpt-5-mini

GPT-5 Mini is OpenAI's fast, cost-effective model. It delivers solid general-purpose performance with low latency and a generous 400K context window.

When to use: Everyday chat and Q&A, rapid prototyping, high-throughput pipelines, and tasks where speed and cost matter more than peak reasoning capability.

When to consider alternatives: For tasks requiring deep analysis or flagship-level quality, use gpt-5.2. For coding tasks, prefer gpt-5.2-codex.

gpt-5.2

GPT-5.2 is OpenAI's flagship model. It represents the best overall quality available from OpenAI, with strong performance across writing, reasoning, analysis, and code.

When to use: Complex analysis, long-form content generation, multi-step reasoning, tasks where output quality is the primary concern, and scenarios where you want OpenAI's best without a coding-specific specialization.

When to consider alternatives: If cost or speed is a priority, use gpt-5-mini. If the task is primarily code generation or agentic tool use, gpt-5.2-codex is optimized for that.

gpt-5.2-codex

GPT-5.2 Codex is a variant of GPT-5.2 specifically optimized for code generation and agentic workflows. It excels at producing structured outputs, calling tools reliably, and executing multi-step coding tasks.

When to use: Code generation, code review, refactoring, debugging, building agentic pipelines, and any workflow where the model needs to produce tool calls or structured JSON outputs with high reliability.

When to consider alternatives: For general conversation or writing tasks unrelated to code, the base gpt-5.2 model is a better fit. For budget-conscious coding tasks, qwen3-coder on OpenRouter offers a lower-cost alternative.

cmo-agent (Custom Fine-Tuned Model)

CMO Agent is Majestix AI's proprietary fine-tuned model — a custom OpenAI model trained specifically to function as a world-class Chief Marketing Officer. This is not a standard provider model; it is exclusive to the Majestix AI Inference Hub.

What Makes It Custom

The CMO Agent is fine-tuned on marketing-specific training data covering:

• Campaign strategy — go-to-market plans, launch playbooks, growth strategy

• Copywriting — ad copy, email sequences, landing pages, social media posts

• SEO & content — keyword strategy, content calendars, blog outlines

• Brand & positioning — messaging frameworks, competitive positioning, brand voice

• Analytics — KPI frameworks, funnel analysis, conversion optimization

• Advertising — paid media strategy, audience targeting, creative briefs

Strengths

The model is tagged with these specializations in the routing system: marketing, campaigns, seo, email, ads, social media, analytics, content, branding, brand, launch, landing page, kpi, funnel, copywriting, promotion, conversion, audience, advertising, cmo, go-to-market, growth, positioning, messaging, creative, strategy.

When the auto-router detects a marketing-related query, it automatically routes to CMO Agent.

Specifications

Property	Value
Provider	OpenAI (fine-tuned)
Context Window	128K tokens
Max Output	16,384 tokens
Category	Custom
Plan Required	Guru or Pro
Input Cost	$0.30 / 1M tokens
Output Cost	$1.20 / 1M tokens

When to Use

• Marketing strategy documents and campaign plans

• Email campaign sequences and drip flows

• Social media content calendars and post copy

• Ad copy for Google, Meta, LinkedIn, Twitter

• SEO content planning and keyword strategy

• Brand messaging and positioning frameworks

• Landing page copy and conversion optimization

• Go-to-market launch playbooks

When to Consider Alternatives

• General-purpose tasks: Use gpt-5.2 or claude-sonnet — CMO Agent's fine-tuning creates strong marketing-oriented opinions that are not ideal for non-marketing tasks

• Code generation: Use gpt-5.2-codex or qwen3-coder

• Large context needs: CMO Agent has a 128K window vs 400K for the GPT-5 family

Behavioral Notes

Because this is a fine-tuned model, it exhibits intentional stylistic preferences:

• Responses default to rich Markdown formatting with headers, bullet points, and tables

• It tends to structure outputs as actionable marketing deliverables rather than conversational text

• It may provide opinionated recommendations based on marketing best practices

• It excels at producing structured frameworks (AIDA, PAS, Before/After/Bridge) without being prompted

This behavior is by design — the model is built to produce production-ready marketing output, not generic conversation.

gpt-5-image

GPT-5 Image generates images from text prompts using OpenAI's image generation capabilities.

When to use: Creating illustrations, concept art, marketing visuals, UI mockups, and any scenario where you need to generate images from natural language descriptions.

When to consider alternatives: For alternative image styles or if you need Google's image generation quality, consider gemini-3-image or seedream-4.5.

Note: Image generation models use flat per-image credit pricing rather than per-token billing.

Context Window Advantage

The GPT-5 family offers 400K token context windows, which is double the 200K available in Claude models. This makes them particularly useful when you need to:

• Process very long documents (contracts, research papers, codebases) in a single request.

• Maintain extended multi-turn conversations without truncating history.

• Include large amounts of reference material alongside your prompt.

The cmo-agent model has a smaller 128K context window due to its fine-tuning configuration. Plan input sizes accordingly when using this model.

Pricing Tier

GPT-5 Mini is one of the more economical options among major-provider models. GPT-5.2 and GPT-5.2 Codex are priced at the premium tier. CMO Agent is priced comparably to GPT-5.2 due to its fine-tuned specialization. All costs are handled through the platform's credit reservation system.