Urgent.News

What's breaking now, across thousands of outlets.

AI

The AI Credit Resale Economy

The AI credit resale economy has emerged as startups and established companies alike begin reselling unused credits at discounted rates. This market phenomenon came to light after a journalist, following a piece on token relay markets, started receiving offers for Anthropic tokens at steep discounts. Founders across various forums and groups were also experiencing similar inbound emails from potential buyers and sellers.

The brokers, who act as intermediaries between providers and customers, are profiting from the supply of credits, which can range from $100k per day to other denominations.

Several websites, such as AI Credits and AICreditMart, have sprung up to facilitate credit marketplaces, offering discounts through bulk purchases. These sites position themselves as providing access to major cloud and inference providers, with easy onboarding processes and various delivery methods. Other brokers, like CheapCredits, claim to offer discounts through bulk pricing, but it's unclear how they acquire the supply.

Telegram channels and startup groups have also become hubs for credit brokering activity, with some channels being relatively active. The total number of credits being offered across these platforms is estimated to be in the tens of millions. However, the ease of offering these credits also provides a breeding ground for abuse. As companies become more aware of the costs associated with this market, crackdowns on abuse are likely to follow.

Written by urgent.news from Hacker News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Also reported by 1 other outlet

Read the original at vectoral.com →

More in AI

I published our agent-security benchmark, including the attacks we fail to catch

Every company building AI agent security publishes a detection rate. We were doing it. The problem is that none of these numbers are checkable.

  • Published agent-security benchmark with 497 attacks and 1,172 benign samples
  • Detection engine achieves 99.8% detection rate, 0.09% false positive rate
  • Failures include pretext opener attack and Stack Overflow benign sample

The Official DeepSWE Number Is Footnoted to DeepSeek Harness

On August 13, DeepSeek shipped four things in one day. V4-Pro left preview. Peak and off-peak API prices start at 16:00 UTC on August 16.

  • DeepSeek released four updates on August 13
  • DeepSWE number 62.7 for DeepSeek-V4-Pro-0813 model
  • MIT-licensed deepseek-harness available for preview

More from Sunday 16 August →