OpenAI
OpenAI3d ago
Startups

Build Hour: Valuemaxxing with GPT-5.6

55 min video5 key momentsWatch original
TL;DR

OpenAI shifts the AI optimization mindset from token maxing to value maxing—measuring AI's worth by business outcomes, not consumption.

Key Insights

1

Budget burned in monthsToken maxing burned through annual AI budgets in months; value maxing flips the metric to business outcomes like time saved, quality improved, or work completed.

2

90% cost reductionPrompt caching alone delivered 90% input cost reduction in OpenAI's demo; Ploy saw 89% savings on first messages by caching system prompts and tools across multiple chats.

3

82% fewer tokensContext compaction cut Ploy's token spend by 82%—from 24,000 input tokens down to 4,000—without changing outputs, saving roughly $37,000 per year on web search tool optimization alone.

4

Three-tier model familyGPT-5.6 introduced three models (Soul, Terra, Luna) at different intelligence/cost tiers; Soul at medium reasoning handles most daily work without needing extra high mode.

5

24% input token savingsProgrammatic tool calling executes code in a JavaScript sandbox instead of reasoning through it, saving 24% of input tokens by cutting one full model turn.

6

Cache breaks kill savingsBreaking the KV cache by updating tool schemas mid-loop ruins cost efficiency; Ploy achieved 45% tool schema reduction by loading niche tools on-demand only when needed.

Want this for every new video OpenAI posts? Brevyd summarizes each upload automatically, the morning it drops.

Deep Dive

From Token Maxing to Value Maxing

Charlie opens by defining token maxing as the now-outdated practice of measuring AI adoption by consumption—how many tokens burned, prompts sent, or agents running. Some companies even built internal leaderboards to track daily token usage. By mid-2024, reality hit: enterprises accidentally torched annual budgets in months and started throttling spend. CEOs admitted the token leaderboards were misguided. Value maxing flips this entirely: measure AI by what it accomplishes—work completed, time saved, quality improved—not raw consumption. The pivot hinges on a single question: if you doubled token spend tomorrow, how would you know it was worth it? This requires defining outcomes first, identifying workflows that lead to good outputs, and building evals to prove quality is actually improving. Without evals—clear definitions of what good looks like—you're flying blind.

Coding Value Maxing and GPT-5.6 Models

Charlie walks through value maxing for developers using Codex. The GPT-5.6 family includes three models: Soul (flagship for complex tasks), Terra (balanced intelligence and cost), and Luna (high-volume, cost-focused). Deep Suey benchmarks show 5.6 delivers the most performance per token spent. Charlie's rule: start with Soul on medium reasoning, not extra high. Most daily work runs fine there; only tune up for sophisticated workloads. The demo shows a visual matrix comparing models and reasoning levels side-by-side. Luna at default looks crude; Luna with more reasoning becomes respectable. Sol starts high-fidelity and gets better with more reasoning. The key insight: sometimes a less intelligent model with more reasoning tokens matches a smarter model's performance at lower total cost. Think in terms of cost per task to completion, not cost per token. Charlie also covers three tactical tips: consider trading tokens for time via fast mode, use auto approval to avoid friction while staying safe, and audit agent instructions—models have gotten so capable that six-month-old prompts are often bloated and need trimming.

API Features for Value Maxing

Charlie introduces five API features designed to save tokens and time. Programmatic tool calling gives models a JavaScript sandbox to execute code, eliminating reasoning overhead and round trips—saves 24% of input tokens by cutting one full model turn. Prompt caching specifies which parts of your prompts stay static; demo showed 90% input cost reduction on a 5,500-token prompt, though latency hit a couple seconds. Smart prompt design places dynamic info (like today's date) at the end to preserve cache hits on the static bulk. Persistent reasoning exposes the ability to save reasoning turns across messages, boosting both performance and cache efficiency. Context compaction runs automatically or manually to reduce long conversation histories—the demo showed 82% input token savings (24K down to 4K) for the same output, though compaction itself took 30 seconds. Charlie's emphasis: these aren't silver bullets; they work best on clear use cases.

Ploy's Production Wins with Caching and Tool Design

Lorenzo from Ploy dives deep into caching strategy, the most impactful but overlooked optimization. When agents loop through tool calls, context accumulates: ten tools across ten steps creates quadratic token processing. Prompt caching reduces that by 10x because embeddings from processed tokens are stored. But caching breaks if you modify the context window mid-loop—e.g., dynamically updating tool schemas. Ploy solved this by splitting tools into always-on (used 70% of sessions) and on-demand (niche tools appended only when needed), achieving 45% tool schema reduction and a 33% cost drop in evals. Ploy also sets cache breakpoints: after system prompt and tools, and after workspace memory. This let them achieve 89% cost reduction on first messages in new chats, cutting overall production token spend by 5%. Lorenzo shares three more wins: batching independent tool calls into one step saves reprocessing tokens; designing tools to do multiple actions per call saves output tokens (14% cost reduction with same pass rate); and switching to Exa's highlights feature in web search dropped output by 70%, saving ~$37K annually. The core lesson: audit traces first to find bloated tool outputs and unnecessary work, then apply caching on top.

Founder Perspective and Q&A Insights

Bryant from Ploy explains why startups must stress-test every new model against real evals reflecting actual use cases—marketing strategy, long instructions, page redesign, WordPress imports. As a second-time founder, he admits his prior experiences need unlearning; AI moves so fast that best practices from last year are stale. The Q&A covers critical nuances: Ultra mode in ChatGPT/Codex is Max reasoning plus system prompt modifications that push models toward sub-agents; it's extremely token-intensive and only makes sense for heavy-hitter tasks, not checking email. Sub-agents receive only context they need, not the full chat history. On agent depth, both speakers prefer staying shallow—two layers max—because delegation gets lossy; newer models like GPT-5.6 Soul have much better orchestration abilities than prior versions. Compaction versus new chats: Lorenzo no longer thinks about compaction for routine work and stays in one thread for weeks; new chats only make sense when your setup is heavily pre-loaded with context that's irrelevant to the new task. Overly relying on caching backfires if you haven't first audited traces and eliminated redundant tool calls and outputs—do the 80% of basic work first, then layer in caching.

Takeaways

  • Define and track evals for your AI workflows—what does good output look like and how do you measure it?
  • Start with Soul on medium reasoning for Codex work; only tune up for genuinely complex tasks.
  • Audit your tool outputs and agent loops for bloat before enabling caching; caching only works on stable context windows.
  • Batch independent tool calls into single steps and design multi-action tools to cut redundant token reprocessing.

Key moments

2:48Token Maxing to Value Maxing Pivot

Token maxing is measuring progress by how much AI you're using, how many tokens you're burning, how many prompts you're sending. Value maxing is measuring progress not by how much AI you're using, but what AI is actually helping you to accomplish.

8:15GPT-5.6 Three-Model Strategy

Soul is our frontier model, our flagship model for the most complex coding and professional tasks. Terra is a really great daily driver and good for more balanced work where intelligence and cost or latency both matter. Luna is great for high-volume workloads where the intelligence level is not as demanding but you really do care about cost and latency.

16:00Prompt Caching Demo Results

The input costs were 90% lower when it comes to using the cache. If you've got like a massive system prompt that you want to cache, this is a really great way to save on cost.

22:00Ploy's Cache Breakpoint Strategy

If you set a breakpoint to cache the system prompt and the tools, this now gets saved for every single chat afterwards. The first message if you're able to cache the system prompt and the tool calls, we saw an 89% reduction in the first message in a new chat.

26:00Context Compaction Impact

We've got 82% fewer input tokens for the same task. Without compaction we were sending 24,000 input tokens and then after it was a little over 4,000.

You just read one. Brevyd does this for every upload.

Follow OpenAI and every new video comes back as a summary like this, in your morning briefing. No watching required.