Here is the cost architecture that matters, in the order you should implement it.

Does Reasoning Effort Actually Change the Bill?

Substantially. Higher effort means more internal tokens generated before the visible answer, and those tokens are billed. A pipeline that defaults to high effort on every request is paying for reasoning that most requests do not need.

Practical settings:

• Extraction and classification: lowest effort. There is no reasoning to do.
• Summarization and rewriting: low to medium.
• Debugging and analysis: medium, escalating to high only on retry.
• Formal math and planning: high, but route to the flagship model only here.

Start every pipeline at the lowest effort that passes your quality bar. Raise it per endpoint, based on measured failures, not on a hunch.

How Should You Route Between Models?

Build a two-tier router from day one. Send all requests to the mini tier first with a validation step. On schema failure or low confidence, escalate the same request to the flagship.

Most teams stop at the second row and never implement caching. Since system prompts and tool schemas are usually identical across calls, cacheable prefixes often make up the majority of input tokens. GPT-5.2 bills cached input at a steep discount, and the only requirement is that the prefix stay byte-identical.

Why Does Structured Output Save So Much?

Free-form prose is where output tokens explode. When you request a typed schema, the model returns the fields and stops — no preamble, no restating the question, no closing summary. On extraction workloads this routinely halves output tokens.

It also removes an entire class of parsing bugs. A response that always matches your schema is worth more than a slightly smarter response you have to clean up.

What Should You Instrument?

Log four numbers per call: input tokens, cached input tokens, output tokens, and latency, each tagged with the endpoint and model. Almost every team that does this finds one endpoint responsible for the majority of spend.

Watch the output-token column especially. Output is the expensive side, and a chatty summarizer with no length cap is the single most common cause of a surprising bill.

What Are the Traps?

Aggressive caching is not free — if your prefix genuinely varies per user, you are restructuring code for zero benefit. Terse prompts can also degrade quality on tasks that need context; measure quality after each optimization, not just cost. And escalation logic can loop if a validator is too strict, doubling spend on requests that were fine the first time. Cap escalation at one retry.

Quick Summary

• Reasoning effort is a per-request cost dial — start at the lowest that passes.
• Route routine traffic to the mini tier and escalate once on failure.
• Cache identical system prompts and tool schemas.
• Use structured output to cut output tokens roughly in half.
• Log tokens per endpoint and find the expensive one before optimizing.

The practical skill with GPT-5.2 is not prompting harder. It is picking the right effort level and the right tier for the job in front of you.

What Is GPT-5.2?

It is the top text model in OpenAI’s lineup, with a large context window and adjustable reasoning effort. Higher effort means the model works through a problem in more internal steps before answering — better on hard analysis, slower and more expensive on everything else.

It handles code, prose, data interpretation, and long-document synthesis. The mini variant covers most routine work at a fraction of the cost, and knowing when to switch is where the savings live.

How Do You Choose an Effort Level?

Default to low or medium. High effort on a summarization task burns tokens and latency for no quality gain — the model is not thinking harder about a task that has no hard part.

How Do You Prompt It for Real Work?

Three habits matter more than clever phrasing:

• Give a role and an audience. “You are reviewing this contract for a non-lawyer operations manager” produces usable output. Without an audience, you get generic coverage.
• Specify the output format. The model will invent a format otherwise, and re-formatting its answer wastes your time.
• Ask for the reasoning only when you need it. “Show your working” doubles output length. Ask for the conclusion and a one-line rationale for routine tasks.

How Much Does GPT-5.2 Cost?

ChatGPT plans start free and step up through paid tiers; the API bills per million input and output tokens, with output costing several times input. At scale, the tier choice dwarfs the prompt choice.

If most of your calls are mini-tier, your bill drops by roughly an order of magnitude. Route deliberately. GPT-5.2 earns its price on hard problems, not on volume.

Where Does It Still Disappoint?

Hallucinated specifics — dates, citations, version numbers — remain the top failure. It also over-hedges on questions with a clear answer, and its default prose runs long unless you cap it. For work that must be factually exact, treat it as a drafting aid and verify every number.

Speed is the other tradeoff. At high effort, responses can take noticeably longer than mini-tier answers, which matters in interactive chat but not in batch jobs.

Who Should Use It?

Analysts, writers, and developers doing genuinely hard work where a better answer is worth the latency. If your tasks are routine — summarizing, formatting, classifying — you are paying a premium for capability you are not using.

Quick Summary

• GPT-5.2 is OpenAI’s flagship text model with adjustable reasoning effort.
• Default to low or medium effort; reserve high for hard analysis.
• Route routine volume to the mini tier for a much lower bill.
• Always specify audience and output format.
• Verify every specific fact — hallucinated details remain the top failure.