# So I've been using gpt-5.6 for awhile... > [!question] Lingering Questions > - Do the reported $180-240K inference spend and the "2-4 hours per native rewrite" claims hold up under any independently reproduced test, or are they early-access-hype numbers? > - What did Theo's promised follow-up review/Fable comparison/pricing breakdown on gpt-5.6-sol actually find? > [!tip] Article Ideas > - A short checklist piece: "what to verify before acting on an early-access model review" — using this video's unreproduced spend and rewrite claims as the worked example. > [!link] Cross Reference > - [[AI Cost & Tokenomics MOC]] — self-reported inference-spend claims are a tokenomics data point worth skepticism > - [[AI & the Future of Knowledge Work MOC]] — native-app rewrite speed claims bear on AI-assisted development capability > [!quote] Vocabulary > - none specific — general early-access product-review vocabulary (native rewrite, inference spend) ## Bib capsule Theo (t3.gg), "So I've been using gpt-5.6 for awhile...," YouTube, 2026. https://www.youtube.com/watch?v=mHG7K7QmQyU ## Author's thesis (Adler Q1+Q2) Early access to gpt-5.6 produced large, self-reported gains — an estimated $180-240K of inference usage over roughly six weeks, two from-scratch native rewrites of a React Native app (AppKit/Swift and SwiftUI separately) completed in 2-4 hours each, and prompts on which gpt-5.5 reportedly got "confused and lost" succeeding cleanly on 5.6. None of these claims come with a controlled methodology or independent reproduction — they're presented as a creator's lived-usage account, which sets up the video's own implicit tension between genuine capability signal and unverified vendor-adjacent enthusiasm. ## Key arguments (≤3) 1. Reported spend of $180-240K in inference over ~6 weeks establishes the scale of early-access usage — a credibility-and-skepticism-setting claim, not itself evidence of quality. 2. Two from-scratch native rewrites (AppKit/Swift, SwiftUI) completed in 2-4 hours each is the concrete, checkable capability claim in an otherwise anecdotal narrative. 3. A direct model-to-model comparison (5.6 succeeding where 5.5 "got confused and lost" on identical prompts) is offered with no controlled methodology. ## Claim atoms ### Reported spend of $180-240K in inference over a month and a half > [!quote] ID:rw_mHG7K7QmQyU_1 > "Depending on how you estimate my usage, I did between $180,000 and $240,000 of inference in my month and a halfish of access." Opening claim establishing the scale of early-access usage; sets up the credibility/skepticism question explored later in the video. ### Rebuilt the React Native app twice, natively, in a few hours each > [!quote] ID:rw_mHG7K7QmQyU_2 > "I had it rebuild the entire React Native app using AppKit and Swift, as well as Swift UI separately. So that's two native rewrites from scratch... it did that in like 2 to 4 hours for each of them." Concrete claimed capability demonstration for mobile-native code generation — the specific, checkable claim in an otherwise promotional narrative. ### 5.6 succeeded where 5.5 got "confused and lost" on the same prompts > [!quote] ID:rw_mHG7K7QmQyU_3 > "5.6 blasted through that. I tried the same prompts later on with 5.5 and it just got confused and lost." Direct model-to-model comparison claim, used as evidence of a capability jump; anecdotal, with no controlled methodology given. ## Permanent candidates - **Vendor-provided usage anecdotes need independent verification before adoption decisions** — strong, portable candidate; promoted below. ## Processing checklist - [ ] Watch for Theo's promised follow-up review/Fable comparison/pricing breakdown on gpt-5.6-sol before drawing conclusions - [x] Connect into the relevant MOC — [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] ## Links - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — early-access AI coding-capability claims and their verification burden - [So I've Been Using GPT-5.6 for Awhile...](3.%20Resources/Readwise/Articles/So%20I've%20Been%20Using%20GPT-5.6%20for%20Awhile....md) — full source transcript and deep-read notes