BLUN King

Maximum AI capability for complex projects

BLUN King is the most capable model in the BLUN family, developed for work where an ordinary chatbot runs out of depth. It grasps complex goals, breaks them down into sensible steps and holds the thread even as a project grows. It can analyse, plan, write code, work with tools and direct several specialised agents towards a single result.

King does more than answer an individual question. It can turn a goal into a complete piece of work: first assess the starting position, then identify risks and dependencies, draw up a plan, carry out the necessary steps and check the result. That makes King particularly suited to long-running agents and to projects in which each decision builds on the last.

With a context window of 1,000,000 tokens, King can take account of exceptional volumes of information within a single coherent working context. That includes complete documentation sets, very long conversations, technical specifications, extensive codebases and large quantities of project material. Key information therefore does not have to be supplied again after every step. King can keep track of connections, decisions and dependencies even across very large projects.

Context window by tier

• King on the free tier: up to 1,000,000 tokens
• King on a paid tier: up to 1,000,000 tokens

With King the maximum context window is the same size on both kinds of tier. A free King conversation can therefore draw on fundamentally the same coherent body of information as King on a paid tier. Any difference between the free tier and a paid offering has to be explained through other tier characteristics — included usage, speed, prioritisation or additional platform capabilities — and not through an artificially reduced context window.

Thinking complex work through properly

King is the model for situations in which a quick, surface-level answer is not enough. It checks preconditions, recognises contradictions and takes account of how a decision affects later steps. Where the information given is incomplete, it can identify what is missing, flag the assumptions it is making and ask specifically for what it needs.

This matters above all in strategic decisions, technical architectures and processes a business depends on. King considers not only the desired result but the route to it: what has to happen first? Which dependencies exist? Where can the plan fail? And how can it be demonstrated at the end that the work was genuinely completed?

Demanding software development

King can understand complex software projects and work on them across multiple files, components and systems. It analyses existing code, recognises how the pieces relate and can plan new functionality so that it fits the architecture already in place.

Typical work includes:

• planning complete software architectures
• developing new features across several components
• analysing and modernising larger codebases
• tracking down faults with several possible causes
• designing databases, APIs and interfaces
• reviewing security and quality
• writing technical documentation and migration plans
• coordinating several coding agents

King is the right choice above all where the point is not merely to produce code but to understand the technical problem as a whole. It can hold requirements against the current state, identify risks before implementation begins and afterwards check whether the solution genuinely meets the original requirements.

Built for long-running agents

A capable agent has to do more than produce text. It has to take decisions, choose tools, assess intermediate results and adjust its approach when something goes wrong. King is designed for precisely this kind of work.

King can break an overarching goal down into individual tasks, work with suitable tools and bring in specialised models. Queen can draft a professional text or a code component, Prince can pre-sort large volumes of data, Voice can process a conversation and Imagine can produce the images required. Throughout, King holds the overall goal together and decides which step makes sense next.

This makes King suitable for:

• autonomous project agents
• complex research and analysis processes
• software development across multiple stages
• internal company agents with a range of tools
• long-running work with intermediate checks
• processes involving several specialised subagents
• projects in which results have to be verifiably checked

Long documents and extensive project context

King can do more than summarise long documents: it can compare them with one another and place them in a wider context. It can bring together requirements from different sources, recognise where they diverge and derive concrete decisions or packages of work from them.

Examples include technical specifications, contracts, project documentation, research material, minutes, policies and larger collections of internal information. From these King can produce a clear overview, a risk analysis, an implementation plan or a structured basis for a decision.

Where King sits in the BLUN model family

King is the highest capability tier within the BLUN model family. The model is intended for maximum depth of thought, difficult reasoning, long-running agents and complex professional work.

King should be chosen when the task has to be thought through as reliably as possible and the additional processing time is justified compared with a quick standard answer. For clearly defined day-to-day tasks, Queen is usually more efficient. For very short or frequently recurring tasks, Prince is the more economical choice.

Typical tasks for King

• “Analyse our entire platform and develop a modernisation plan that will hold up.”
• “Plan this software in full, identify the technical risks and coordinate the build.”
• “Compare these contracts, find the contradictions and produce a basis for deciding.”
• “Investigate this fault across frontend, backend and database, and prove the cause.”
• “Turn this business goal into an agent process I can run, with clear controls.”
• “Carry this project through several stages and check every result at the end.”

When King is the right choice

Use King when:

• the problem is complex or not yet fully structured,
• several decisions depend on one another,
• large amounts of context have to be taken into account,
• tools or several agents need to be coordinated,
• high-quality software development is required,
• a wrong or incomplete result would be expensive to put right,
• the task has to be followed through across many steps.

When another model makes more sense

For professional tasks with a clearer scope, Queen usually offers the better balance of capability, speed and cost. Prince is the better fit when thousands of short requests, classifications or standard processes have to be handled. Voice and Imagine are used when speech or images are the focus.