Os negacionistas básicamente impiden que se regule porque as AI é o mal e os tokenmaxing o ven como a solución a librarse deses traballadores que tanto cobran. Así está o tema.
November 3, 2026 at 8:24 AM
Os negacionistas básicamente impiden que se regule porque as AI é o mal e os tokenmaxing o ven como a solución a librarse deses traballadores que tanto cobran. Así está o tema.
Q2 was flooded with tokenmaxing by companies. Q3 numbers will explain more/better.
October 26, 2026 at 2:25 PM
Q2 was flooded with tokenmaxing by companies. Q3 numbers will explain more/better.
Consumption vs subscription revenues
tokenmaxing
tokenmaxing
October 26, 2026 at 2:52 PM
Consumption vs subscription revenues
tokenmaxing
tokenmaxing
Tokenmaxing consumption revenues
October 26, 2026 at 3:01 PM
Tokenmaxing consumption revenues
How much of this is due to tokenmaxing?
How much of this will vaporize when token maxxing is gone?
How much of this will vaporize when token maxxing is gone?
October 21, 2026 at 7:36 AM
How much of this is due to tokenmaxing?
How much of this will vaporize when token maxxing is gone?
How much of this will vaporize when token maxxing is gone?
A stark reversal from the "tokenmaxing" days, we're now increasingly seeing firms talk about efficiency and ways to get the tasks done in the fewest tokens possible so they don't get to the end of their budget before the end of their month.
October 21, 2026 at 2:55 PM
A stark reversal from the "tokenmaxing" days, we're now increasingly seeing firms talk about efficiency and ways to get the tasks done in the fewest tokens possible so they don't get to the end of their budget before the end of their month.
On a different note, What if Weinersmith INC used “ARR”….not “Annualized Return Rate” but “Annualized Run Rate” (where you select a month or day of your choice and multiply it to an estimated year revenue) Especially after the end of “tokenmaxing“.
Would that not be cause of concern?
Would that not be cause of concern?
October 21, 2026 at 2:55 PM
On a different note, What if Weinersmith INC used “ARR”….not “Annualized Return Rate” but “Annualized Run Rate” (where you select a month or day of your choice and multiply it to an estimated year revenue) Especially after the end of “tokenmaxing“.
Would that not be cause of concern?
Would that not be cause of concern?
Love how the try to soften it by first saying "Red Hot Growth" when that's entirely from their enterprise customers tokenmaxing Q1-2 before the token based billing had companies realizing they were spending $10Ms more than they realized and switched to hard rations. 2nd half '26 will be barren.
October 21, 2026 at 12:39 AM
Love how the try to soften it by first saying "Red Hot Growth" when that's entirely from their enterprise customers tokenmaxing Q1-2 before the token based billing had companies realizing they were spending $10Ms more than they realized and switched to hard rations. 2nd half '26 will be barren.
Regenwetter oder wies bei uns zu Hause heisst: Tokenmaxing
October 17, 2026 at 4:07 PM
Regenwetter oder wies bei uns zu Hause heisst: Tokenmaxing
People stopped tokenmaxing.
October 15, 2026 at 3:16 AM
People stopped tokenmaxing.
nei, en ég myndi halda að þetta væri svipað í öðrum greinum, að þú þarft manneskju til að stýra þessu og að fyrirtæki væru almennt frekar brunnin eftir þessa tokenmaxing vitleysu í fyrra
October 13, 2026 at 4:42 PM
nei, en ég myndi halda að þetta væri svipað í öðrum greinum, að þú þarft manneskju til að stýra þessu og að fyrirtæki væru almennt frekar brunnin eftir þessa tokenmaxing vitleysu í fyrra
“Necati, how can I burn more tokens?”
Someone asked me this recently.
I had to pause. Some companies confuse tokenmaxing with efficiency. If you measure AI activity instead of progress, you are not leading an AI-first engineering team. You are approving a larger API bill.
Someone asked me this recently.
I had to pause. Some companies confuse tokenmaxing with efficiency. If you measure AI activity instead of progress, you are not leading an AI-first engineering team. You are approving a larger API bill.
October 13, 2026 at 7:30 PM
“Necati, how can I burn more tokens?”
Someone asked me this recently.
I had to pause. Some companies confuse tokenmaxing with efficiency. If you measure AI activity instead of progress, you are not leading an AI-first engineering team. You are approving a larger API bill.
Someone asked me this recently.
I had to pause. Some companies confuse tokenmaxing with efficiency. If you measure AI activity instead of progress, you are not leading an AI-first engineering team. You are approving a larger API bill.
Of course, the tokenmaxing was right before they started hiking up token prices.
But the actual trick is to read programming theory books and literally distill their knowledge into extension files the LLM can read and apply to the code it generates.
But the actual trick is to read programming theory books and literally distill their knowledge into extension files the LLM can read and apply to the code it generates.
October 9, 2026 at 12:06 AM
Of course, the tokenmaxing was right before they started hiking up token prices.
But the actual trick is to read programming theory books and literally distill their knowledge into extension files the LLM can read and apply to the code it generates.
But the actual trick is to read programming theory books and literally distill their knowledge into extension files the LLM can read and apply to the code it generates.
I wonder if we've hit peak tokenmaxing. You can get good enough without paying the Claude surcharge recently.
September 29, 2026 at 2:30 AM
I wonder if we've hit peak tokenmaxing. You can get good enough without paying the Claude surcharge recently.
not for my core work at least
i could defs make some useful shit happening with tokenmaxing IFF i had finished the tools i’m presently about to start working on
but i don’t! so until i do
gonna work a bit more hands on i think!
i could defs make some useful shit happening with tokenmaxing IFF i had finished the tools i’m presently about to start working on
but i don’t! so until i do
gonna work a bit more hands on i think!
September 16, 2026 at 5:08 PM
not for my core work at least
i could defs make some useful shit happening with tokenmaxing IFF i had finished the tools i’m presently about to start working on
but i don’t! so until i do
gonna work a bit more hands on i think!
i could defs make some useful shit happening with tokenmaxing IFF i had finished the tools i’m presently about to start working on
but i don’t! so until i do
gonna work a bit more hands on i think!
Their revenue also does in fact keep doubling every few months.
Like the stuff about how revenue going up was just because over expending on tokens due to tokenmaxing and soon it would go down wasn't true(or well was never concrete enough to be falsifiable), it just kept going up.
Like the stuff about how revenue going up was just because over expending on tokens due to tokenmaxing and soon it would go down wasn't true(or well was never concrete enough to be falsifiable), it just kept going up.
September 10, 2026 at 3:06 AM
Their revenue also does in fact keep doubling every few months.
Like the stuff about how revenue going up was just because over expending on tokens due to tokenmaxing and soon it would go down wasn't true(or well was never concrete enough to be falsifiable), it just kept going up.
Like the stuff about how revenue going up was just because over expending on tokens due to tokenmaxing and soon it would go down wasn't true(or well was never concrete enough to be falsifiable), it just kept going up.
Yeah, anthropic's move that shocked people out of the very silly tokenmaxing fad.
Given what people are finding with the quality of open weight competitors, the effects of harness choice on performance and lower price inference from outside the US - some folk stateside must be getting quite nervous.
Given what people are finding with the quality of open weight competitors, the effects of harness choice on performance and lower price inference from outside the US - some folk stateside must be getting quite nervous.
August 22, 2026 at 10:17 AM
Yeah, anthropic's move that shocked people out of the very silly tokenmaxing fad.
Given what people are finding with the quality of open weight competitors, the effects of harness choice on performance and lower price inference from outside the US - some folk stateside must be getting quite nervous.
Given what people are finding with the quality of open weight competitors, the effects of harness choice on performance and lower price inference from outside the US - some folk stateside must be getting quite nervous.
Darko says, he's not token maxing, do you believe him?
Watch the whole thing on YT
#kiro #kirocrew #tokenmaxing 🧵
Watch the whole thing on YT
#kiro #kirocrew #tokenmaxing 🧵
August 21, 2026 at 1:14 PM
Darko says, he's not token maxing, do you believe him?
Watch the whole thing on YT
#kiro #kirocrew #tokenmaxing 🧵
Watch the whole thing on YT
#kiro #kirocrew #tokenmaxing 🧵
KI-Unternehen seit Jahren: "KI nimmt die kleinen simplen Arbeiten ab, wie ein persönlicher Assistent"
Kritik des Tokenmaxing: "Mitarbeiter nutzen KI für Aufgaben, die sie selber machen könnten, deshalb geben die Firmen zu viel für Token aus"
Kritik des Tokenmaxing: "Mitarbeiter nutzen KI für Aufgaben, die sie selber machen könnten, deshalb geben die Firmen zu viel für Token aus"
August 19, 2026 at 5:08 AM
KI-Unternehen seit Jahren: "KI nimmt die kleinen simplen Arbeiten ab, wie ein persönlicher Assistent"
Kritik des Tokenmaxing: "Mitarbeiter nutzen KI für Aufgaben, die sie selber machen könnten, deshalb geben die Firmen zu viel für Token aus"
Kritik des Tokenmaxing: "Mitarbeiter nutzen KI für Aufgaben, die sie selber machen könnten, deshalb geben die Firmen zu viel für Token aus"
Spot on. It’s the brutal reality of "Tokenmaxing"—where the revenue run rate looks like a feast, but the unit economics can feel like a famine.
NVDA ($227.08) represents those $50 ingredients. With AWS hiking AI capacity prices by ~15–20% recently, that "menu price" is getting eaten alive by…
NVDA ($227.08) represents those $50 ingredients. With AWS hiking AI capacity prices by ~15–20% recently, that "menu price" is getting eaten alive by…
August 17, 2026 at 4:28 PM
Spot on. It’s the brutal reality of "Tokenmaxing"—where the revenue run rate looks like a feast, but the unit economics can feel like a famine.
NVDA ($227.08) represents those $50 ingredients. With AWS hiking AI capacity prices by ~15–20% recently, that "menu price" is getting eaten alive by…
NVDA ($227.08) represents those $50 ingredients. With AWS hiking AI capacity prices by ~15–20% recently, that "menu price" is getting eaten alive by…
Denk dat dit echt nog voor problemen gaat zorgen. Tot nu toe werd alles dermate gesubsidieerd dat veel mensen compute behandelden als kostenloos (cfr tokenmaxing). Die subsidies gaan per definitie wegvallen, is de facto al zo met fable die alleen via API pricing beschikbaar blijft.
July 18, 2026 at 6:45 AM
Denk dat dit echt nog voor problemen gaat zorgen. Tot nu toe werd alles dermate gesubsidieerd dat veel mensen compute behandelden als kostenloos (cfr tokenmaxing). Die subsidies gaan per definitie wegvallen, is de facto al zo met fable die alleen via API pricing beschikbaar blijft.
Notes from Google I/O, including:
- What Sundar Pichai told me about Mythos
- How Google is setting up for what I'm calling the tokenmaxing hangover
- My interviews with the SVPs of Google Search and Devices
sources.news/p/google-tok...
- What Sundar Pichai told me about Mythos
- How Google is setting up for what I'm calling the tokenmaxing hangover
- My interviews with the SVPs of Google Search and Devices
sources.news/p/google-tok...
Google gets ready for the tokenmaxxing hangover
Notes from Google I/O. Also: What Andrej Karpathy is doing at Anthropic, Microsoft's special CEO summit guest, and partnering with Parallel.
sources.news
May 20, 2026 at 6:27 AM
Notes from Google I/O, including:
- What Sundar Pichai told me about Mythos
- How Google is setting up for what I'm calling the tokenmaxing hangover
- My interviews with the SVPs of Google Search and Devices
sources.news/p/google-tok...
- What Sundar Pichai told me about Mythos
- How Google is setting up for what I'm calling the tokenmaxing hangover
- My interviews with the SVPs of Google Search and Devices
sources.news/p/google-tok...
You forgot people are “tokenmaxing”. Efficiency has sailed long ago.
May 29, 2026 at 7:12 AM
You forgot people are “tokenmaxing”. Efficiency has sailed long ago.
#Amazon employees are faking their #AI usage 🤣 because managers are forcing them to use internal AI agent #MeshClaw, so they turn to "Token Maxing".
We support the internal #AIresistance by #TokenMaxing
youtu.be/Mhm1C76GhoU?...
We support the internal #AIresistance by #TokenMaxing
youtu.be/Mhm1C76GhoU?...
Amazon Employees Are Faking Their AI Usage
YouTube video by The WAN Show
youtu.be
May 22, 2026 at 7:50 PM
#Amazon employees are faking their #AI usage 🤣 because managers are forcing them to use internal AI agent #MeshClaw, so they turn to "Token Maxing".
We support the internal #AIresistance by #TokenMaxing
youtu.be/Mhm1C76GhoU?...
We support the internal #AIresistance by #TokenMaxing
youtu.be/Mhm1C76GhoU?...