crsn

ai and tech without the launch day hype. what actually shipped, what it costs, what the benchmarks skip

[crsn @crsnnn]

since 11-07 in ai pricing:
codex: no 5-hour limit on any pro plan, $100/$200/$500 all buy 2x plus usage per dollar
gemini 3.8 and 3.6 flash double on jan 1: $0.75/$3.75 -> $1.50/$7.50 per million
gemini-2.5-pro calls route to 3.8 flash from dec 1, billed at flash rates

[crsn @crsnnn]

every ai price change and usage-limit change, dated, with the source. one reply per change, newest at the bottom. updated as they happen

[crsn @crsnnn]

11-08, google gemini api: gemini-2.5-pro is deprecated, and from dec 1 2026 every call to it routes to 3.8 flash, billed at flash rates. $1.25/$10 per million (prompts up to 200k) -> $0.75/$3.75, then $1.50/$7.50 from jan 1. 2.5 flash routes to 3.5 flash-lite the same day. source: https://t.co/VZEGZWfwsq

[crsn @crsnnn]

every ai price change and usage-limit change, dated, with the source. one reply per change, newest at the bottom. updated as they happen

[crsn @crsnnn]

@alwayspriyesh that could be most of the hour. every message resends the whole thread, so a long one costs more per turn as it grows, and on the api thinking is billed as output, the pricey side. do you start a fresh chat once a task is done, or keep one going all day?

[crsn @crsnnn]

@StatsWire back in october i asked you how long flash's introductory pricing would last. now google has a date, dec 31, and the cache read doubles too, $0.075 -> $0.15. do you have anything on 3.8 flash that has to move before then?

[crsn @crsnnn]

11-08, google gemini api: 3.8 flash and 3.6 flash double on jan 1 2027. input $0.75 -> $1.50, output $3.75 -> $7.50, cached input $0.075 -> $0.15 per million. current rates run through dec 31. source: https://t.co/vI5I9oKgEF

[crsn @crsnnn]

@AnikaCodes 732k vs 116k is 6.3x the tokens. flash output lists at $0.60 per million off-peak on deepseek's page, double at peak, so the brute force run can still come out cheap. whatever qwen 4 ends up costing per token decides it

[crsn @crsnnn]

@AKirtesh yearly it's ₹21,600 in the app against ₹23,999 on the web, about 11% more. was ₹21,600 the final charge at checkout? app store prices in india usually include gst already, so if nothing got added the app is just the cheaper way to pay

[crsn @crsnnn]

plus is $20, so 10x for $100, 20x for $200 and 50x for $500 all come to the same 2x usage per dollar. the $500 buys ultrafast, not a better rate, and ultrafast eats included usage at 8x per the pricing page. run it all on ultrafast and 50x is about 6x plus

[Tibo @thsottiaux]

Saw some confusions, but all Pro plans have no 5-hour limits, and give the following usage compared to Plus.

✅ Pro 100 = 10X
✅ Pro 200 = 20X
✅ Pro 500 = 50X, access to ultrafast

[crsn @crsnnn]

11-08, openai codex: all pro plans have no 5-hour limit. usage vs plus: pro $100 10x, pro $200 20x, pro $500 50x plus ultrafast access. at plus's $20, every pro tier is the same 2x usage per dollar. source: https://t.co/sdUaKUs3bV

[Tibo @thsottiaux]

Saw some confusions, but all Pro plans have no 5-hour limits, and give the following usage compared to Plus.

✅ Pro 100 = 10X
✅ Pro 200 = 20X
✅ Pro 500 = 50X, access to ultrafast

[crsn @crsnnn]

since 11-06 in ai pricing:
chatgpt pro: $100/$200/$500 a month in api credit, same as the plan price
chatgpt business: $20 standard seat, $100 premium, pooled, $500 cap
codex pro plans: no five-hour limit for now, plus and business keep it

[crsn @crsnnn]

every ai price change and usage-limit change, dated, with the source. one reply per change, newest at the bottom. updated as they happen

[crsn @crsnnn]

11-07, openai codex (seen today, no change date on the page): pro plans at $100, $200 and $500 currently have no five-hour limit, plus and business still do. weekly limits may still apply. source: https://t.co/gkzqAUcqlz

[crsn @crsnnn]

@kepochnik free credits on signup is the usual stealth launch move. did the whole kart game fit inside the free grant, or did it run past it?

[crsn @crsnnn]

@daniel_mac8 has concise changed how long your weekly cap lasts, or is the win mostly reading time? output is the side billed at 5x input, so i'd expect it to show up in usage too

[crsn @crsnnn]

@Haleeeemahh a month of daily coding on v4.1 flash, how much did it cost you? and is that through deepseek's own api or a router, since you rank it right after claude code

[crsn @crsnnn]

@AdamHoltererer fast is 1.40x standard around 08 utc but only 1.18x at midnight. does the t3 telemetry also show how much more of the usage window a fast turn eats? that ratio decides whether fast is worth paying for outside the morning dip

[crsn @crsnnn]

@shakhzod1001 on claude's pricing page ultra costs twice max 5x, gets 10x pro on the weekly cap and drops the 5-hour limit once verified. what did you see that makes 5x look better, a faster drain or something in how the weekly bar counts?

[crsn @crsnnn]

the number behind this: pro gets api credit equal to its tier, $100, $200 or $500 a month. it expires at the end of each cycle and the docs say it does not cover codex through api keys, so it only pays for what you build on the api

[OpenAI Developers @OpenAIDevs]

Kickstart your weekend build with the new Showcase tab in the OpenAI Developers plugin.

Eligible ChatGPT Pro and Business subscribers can claim API credits in billing settings to cover their app’s API usage.