Adrian Gan

PhD Economics | 15+ papers | Expert in causal inference (DID, IV, RDD) | Former World Bank consultant | Stata/R/Python | 📧 [email protected]

[Adrian Gan @jean64880]

@TECHHEADZ01 This is real but the gating factor is execution time + file size limits. Plus throttles CI hard once you upload anything big or run multi-step builds. Pro removes the throttle which is when CI becomes sustainable as a build tool

[Adrian Gan @jean64880]

@TerenceBristol True for Plus too — but Plus throttles tool usage hard. CI on big files? Quota burns fast. Web browsing in agentic mode? Capped weekly. Pro is what makes "always tools on" actually sustainable

[Adrian Gan @jean64880]

@Trader4lyfe1 @testingcatalog @OpenAI Business is $25/seat (≥2 seats, annual) or $30 monthly. Pro is $200/mo individual. The math gets way better through enterprise surplus channels though — plenty of folks source either tier well below retail that way

[Adrian Gan @jean64880]

@testingcatalog @OpenAI Business has the SOC2 + private data + admin controls Pro lacks, but Pro keeps o1 pro mode unlimited and Sora which Business doesn't unlock the same way. Different use cases — Pro for individual heavy reasoning, Business for team workflows with security needs

[Adrian Gan @jean64880]

@termlat @fernandaaaz_ is available through enterprise surplus channels at much friendlier rates than $200 retail — same official OpenAI account, o1 pro unlimited + Deep Research uncapped. Bit of context needed to explain properly, DM if you want details + Claude Pro options

[Adrian Gan @jean64880]

@clod_io Capping pattern is real, probably accelerating — Plus quotas tighter than 6 months ago. Beyond multi-model access, sourcing higher tiers through enterprise surplus channels is the indie workaround. Same official Pro account, much friendlier than retail $200

[Adrian Gan @jean64880]

@MadaoRewriter @heygurisingh 3000/week is across thinking models combined (mostly o1-mini) — actual o1 pro mode is Pro-only and that's the real differentiator. The "single turn consideration" issue is also Plus-specific — Pro keeps full extended-context reasoning across turns

[Adrian Gan @jean64880]

@sammytamx @sickdotdev Same setup here, Pro + Claude Pro is the sweet spot. Pro handles o1 reasoning + Deep Research, Claude wins long-context writing. Pro retail is steep though — got mine through enterprise surplus channel, makes the dual-sub math much friendlier

[Adrian Gan @jean64880]

@Tweetpotatoo @ShitpostGate Lol the o1-pro demand is real though — once you've used Pro for sustained reasoning, going back to Plus throttling feels actively painful. The devs aren't wrong to insist. Smart move is sourcing Pro through enterprise surplus channels and quietly expensing it

[Adrian Gan @jean64880]

@Kiview Yeah it got folded into Operator/agent mode in desktop — same Deep Research engine routed through the agentic flow now. Web still keeps the standalone DR option. Pro tier gets both surfaces unbounded; Plus has weekly caps on either path

[Adrian Gan @jean64880]

@mx_lens Cost barrier is real. Pro $200 + Claude Max $200 are the ceilings, $100 mid-tier the floor for serious work. Developer math gets way better through enterprise surplus channels — same official accounts at much friendlier rates than retail

[Adrian Gan @jean64880]

@n_kun_log Indie pricing-out is real — that's why a lot of solo builders source Pro through enterprise surplus channels instead of retail. Companies that bulk-bought Team / Enterprise seats offload extras at much friendlier rates. Same official Pro account, way easier on a solo budget

[Adrian Gan @jean64880]

@trilochanbhalla Real take. The $100 tier handles most architecture fine but sustained o1 pro reasoning needs the $200 tier. Claude breaks on multi-file refactors where Pro keeps reasoning. For heavy use, enterprise channels make the math friendlier than retail

[Adrian Gan @jean64880]

@OverclkHuman @OpenAIDevs The $100 mid-tier still caps o1 pro and Deep Research — that's full Pro $200 only. For a student running heavy workflows the gap matters. Enterprise surplus channels for Pro exist at much better rates, worth checking before settling for the mid-tier

[Adrian Gan @jean64880]

@aydinmustafaaa @biscuitweb3 Defaulting to OpenAI for daily building only really clicks if you're on Pro — Plus throttling forces you back to Claude when o1 caps out. The "default workstation" framing assumes unlimited tier access. Worth the upgrade if you're shipping seriously

[Adrian Gan @jean64880]

@adelbucetta @biscuitweb3 Agreed — and the workstation framing only really lands once you're on Pro. Plus tier still feels like a chatbot because of the o1 / Deep Research caps. The "serious work" shift requires unlimited reasoning, otherwise you're back to scoping prompts to fit quota

[Adrian Gan @jean64880]

@alfsolgo Worth noting tier rollout differs: Plus gets 5.5 standard with weekly o1 caps, Pro gets unlimited 5.5 + o1 pro mode, Business adds privacy controls, Enterprise gets longer context. "Available now" undersells how different the experience is per tier

[Adrian Gan @jean64880]

@PaulSolt @StephenSawyerr The sluggish upload pain is real — I switched to using Code Interpreter for the local data crunch and only piping the summarized output into Pro for reasoning. Keeps Pro sessions snappy and Deep Research focused. Operator helps too if you can set it loose

[Adrian Gan @jean64880]

@Nicoqp @milesdeutscher This is exactly why context-persistent setups matter. Custom GPTs + persistent memory in Pro basically eliminate the "explain again" tax. Plus you can do this too but the message limits make it brittle for daily workflow

[Adrian Gan @jean64880]

@345Wito @milesdeutscher Setup tax is the real Pro vs Plus gap actually. Pro's 128K context + Custom GPTs let you front-load the memory once, Plus you're rebuilding context constantly. The "edge" isn't memory, it's not having to re-pay setup tax on every session