GPT-5.6 offers two ways to spend more compute on difficult work: max reasoning and ultra mode. The names suggest a simple ladder—ordinary, max, then ultra—but they represent different execution strategies.
Max gives a GPT-5.6 model more time to reason, explore alternatives, check its work, and revise. Ultra coordinates several agents across parallel workstreams. One goes deeper inside a single reasoning process; the other goes wider by dividing work.
For a creator or content team, that choice becomes useful when it leads to a finished asset. Medux provides practical workflows for AI presenter videos, script-to-speech voiceovers, and image or video cleanup. You can work directly in its browser-based Workshop or follow its MCP tutorials to connect media tasks to Codex or Claude. The examples below show where deeper planning helps and how to move from a brief to a result you can review and download.
OpenAI introduced both with the GPT-5.6 general-availability release on July 9, 2026. Max appears in the documented API reasoning options and is also available in ChatGPT Work and Codex for users with GPT-5.6 access. Ultra is presented as a product mode for eligible ChatGPT Work and Codex plans. For API builders, OpenAI separately offers Multi-Agent orchestration as a beta. It is safer to keep those product and API concepts distinct than to assume “ultra” is an undocumented API reasoning value.
What max reasoning actually changes
The GPT-5.6 model catalog lists reasoning efforts from none through low, medium, high, xhigh, and max. OpenAI says max gives the model more time than xhigh to explore alternatives, run checks, and revise.
That makes max a good fit for problems with a tightly connected chain of thought:
- debugging a failure that crosses several layers of one system;
- reconciling contradictory requirements into one coherent design;
- constructing a proof or technical argument with dependent steps;
- reviewing a high-impact plan for hidden assumptions;
- refining one artifact against many interacting constraints.
In these tasks, splitting the problem can destroy useful context. The result of step three changes how step two should be interpreted, or several constraints must be held together throughout. One reasoning process with more time may outperform a committee of partially informed agents.
Max does not grant new tools, a larger context window, or guaranteed correctness. It changes the inference budget. That normally means greater latency and more reasoning tokens, so the setting should be justified by the expected value of a better result.
Ultra is parallelism, not simply deeper thought
OpenAI describes ultra as its highest-capability setting and says it coordinates multiple agents in parallel. The GPT-5.6 launch post compares a default four-agent ultra setup with a one-agent baseline in several evaluations, and also shows larger configurations for some tests.
The right mental model is an expert team:
- a root agent decomposes the goal;
- focused agents investigate separate workstreams;
- they return findings or artifacts;
- the root reconciles conflicts and produces the final result.
Ultra can help with broad research, exploring different areas of a codebase, comparing several proposals, preparing independent components, or testing multiple hypotheses. It can reduce wall-clock time when those activities genuinely overlap.
It is a poor fit for a single ordered chain, a task where agents would contend over the same file, or work dominated by one slow external operation. Four agents waiting on the same render queue are still waiting on the same render queue.
Max versus ultra at a glance
| Question | Max reasoning | Ultra mode |
|---|---|---|
| Core strategy | More reasoning in one model context | Several agents work in parallel |
| Best task shape | Sequential, tightly coupled | Independent, bounded workstreams |
| Main benefit | Deeper exploration and revision | Faster coverage and specialization |
| Main cost | More reasoning time and tokens | More aggregate tokens and coordination |
| Main failure mode | Overthinking or excessive latency | Duplication, conflict, weak synthesis |
| Shared-state work | Usually simpler | Often problematic |
Neither mode replaces external validation. Max can confidently converge on the wrong assumption. Ultra can produce a polished consensus from several agents that all inherited the same flawed premise.
Availability and naming matter
The API documentation lists max as a reasoning effort for GPT-5.6 Sol, Terra, and Luna. OpenAI’s launch announcement says max can be enabled in ChatGPT Work and Codex wherever the user has GPT-5.6 access.
Ultra has plan-specific product availability: at launch, OpenAI said it was available to Pro and Enterprise users in ChatGPT Work and to Plus and higher plans in Codex. Product eligibility can change, so current plan documentation should be checked before making procurement decisions.
The Responses API Multi-Agent beta exposes model-directed subagents across all GPT-5.6 tiers. Its documented default maximum concurrency is three, which differs from the four-agent ultra configuration described in OpenAI’s launch evaluation. That is another reason not to treat ultra and the API beta as interchangeable labels.
How much does extra compute cost?
OpenAI publishes token prices for Sol, Terra, and Luna, but the relevant cost of max or ultra depends on actual usage. Max may consume more reasoning tokens. Ultra adds contexts, outputs, tool calls, and synthesis across multiple agents. A fixed “ultra costs X times more” claim would be misleading without a specific task and current product billing model.
Measure:
- total input, cached input, reasoning, and output tokens;
- subagent count and duplicated context;
- tool and external-service charges;
- p50 and p95 completion time;
- final acceptance rate;
- human review and correction time.
If max costs twice as much but turns a 70% pass rate into 95% on a high-value task, it may be economical. If ultra saves two minutes on research but triples spend and produces the same decision, it is not.
A decision framework for extra compute
Ask five questions before escalating:
1. Is the task actually difficult?
Do not infer difficulty from prompt length. A long extraction task can be mechanical, while a short instruction may conceal conflicting goals. Run a lower effort first when failure is cheap and detectable.
2. Is the difficulty serial or parallel?
Use max when the solution requires one coherent chain. Use ultra when independent specialists can contribute useful work simultaneously.
3. Can success be evaluated?
Tests, citations, schemas, visual rubrics, and human approval make escalation measurable. Without acceptance criteria, more compute may only produce more persuasive prose.
4. Is an external bottleneck dominant?
Parallel reasoning cannot accelerate a vendor queue, a fixed-duration render, or a human approval. Optimize the critical path instead.
5. What is the failure impact?
High-impact legal, financial, security, identity, or publication actions may warrant stronger reasoning, but they may also require a qualified human. More model compute is not a substitute for authorization.
A tiered strategy beats a permanent toggle
Start routine work at the lowest model and effort level that passes evaluation. Escalate when observable triggers occur: conflicting sources, failed tests, repeated tool errors, an unusually large dependency graph, or a high-risk approval point.
For difficult work, try max before ultra when the problem is tightly coupled. Try multi-agent execution when the work naturally separates. Cache stable instructions, cap agent budgets, and preserve artifacts so an escalation does not repeat expensive external operations.
Put your media plan into practice with Medux
Medux gives creators and teams a place to turn scripts and source media into finished outputs. Its Workshop offers focused browser workflows, while its MCP tutorials explain how to run media tasks from Codex or Claude. That makes the reasoning decision practical: use more compute to improve the brief, script, or edit plan, then submit the prepared inputs to the appropriate Medux workflow.
Consider a training campaign that needs a presenter video, an audio-only narration, and a cleaned-up version of an existing clip. Separate research or script-review workstreams can be prepared in parallel. A tightly connected decision, such as shortening a script while preserving an explanation, can benefit from deeper reasoning. Medux handles the selected media task once those inputs are ready.
Create a professional presenter with Workshop Avatar
Medux Workshop Avatar is a starting point for scripted digital-human videos. Choose a professional role and a supported language, set the presenter options, and enter the words you want spoken. You can also provide your own portrait. Medux brings the speech, presenter image, animation, and lip-sync steps into a workflow that returns the completed video for preview and download.
This is useful for training introductions, product explainers, onboarding, and localized service information. Spend reasoning effort on the message: terminology, tone, audience knowledge, and what should fit into a short spoken script. The professional and language pages then provide a concrete place to create the presenter video.
Find the voice you want with Workshop Voice
Medux Workshop Voice helps you browse AI voices by delivery style, gender, and use case, then turn a script into generated speech. Open a voice profile, adapt its sample script or add your own, and create audio for a narration, explainer, announcement, or other spoken content.
Voice selection affects how a message is heard. A calm instructional delivery and an energetic promotional read serve different briefs. Use the planning stage to choose the tone and simplify sentences for listening; use Medux to generate the speech and review its pronunciation and pacing. This is also a useful route when the deliverable is an audio track rather than a presenter video.
Prepare existing footage with Workshop Remove
Medux Workshop Remove brings together focused image and video cleanup tools for supported backgrounds, watermarks, logos, subtitles, and unwanted objects. Select the tool that matches the material and the change you need, then review the returned output.
For a campaign assembled from existing assets, cleanup can be as useful as generating something new. A team might prepare an approved archived clip for a fresh edit or remove a distracting element from source media. Use files you own or have permission to edit, and check the affected frames before reusing the result.
Automate an audio-driven avatar video through Medux MCP
The Medux audio-driven avatar video tutorial shows how Codex or Claude can pair an avatar source with driving speech audio, call the matching MCP tool, and track the task through to its result. It is a practical next step when you already have the visual source and audio and want the workflow inside your AI client.
Review the returned video's mouth timing and playback. For a larger content package, keep the task ID and source references with the project so a later script or planning revision can reuse the work already completed.
Automate video logo removal through Medux MCP
The Medux video logo removal tutorial covers another useful task: identifying the source video, optional logo reference, and motion mode before submitting a logo-removal job. It explains the Codex and Claude workflows, task tracking, and result review.
This gives a prepared edit plan a specific action. Check that the intended logo is removed throughout the relevant frames and that the returned video plays correctly. Use the browser Workshop for a direct cleanup session or the tutorial when you want to connect the operation to an agent workflow.
Choose the Medux route that fits your deliverable
Start with Avatar when the output should be a speaking presenter, Voice when you need speech audio, and Remove when existing media needs cleanup. Follow the two MCP tutorials when you want Codex or Claude to coordinate the corresponding media task from your project.
Keep uploaded file references, job settings, task IDs, and accepted outputs with the project. A change in reasoning effort should improve the next decision without submitting the same paid media job again. Compare the cost and quality of the complete workflow, including media generation and review, when deciding whether Max or Ultra earns its extra compute.