> ## Content Index
> Fetch the complete content index at: https://www.tao.media/llms.txt
> Use this file to discover other available public pages before exploring further.

# Lium Reports $964K Month as It Tops Bittensor’s Subnet Market Cap Rankings
- URL: https://www.tao.media/lium-reports-964k-month-as-it-tops-bittensors-subnet-market-cap-rankings/
- Published: 2026-09-11T18:39:07.000Z
- Updated: 2026-09-11T18:39:07.000Z
- Description: Lium says its Bittensor GPU marketplace billed $964K last month across 1,187 renters, pushing Subnet 51 to the top spot among non-root subnets by market cap.
- Author: Bart Hillerich
- Tags: Lium, News, Subnets

[Lium](https://lium.io/?ref=tao.media), the decentralized GPU rental marketplace on Bittensor Subnet 51, said it billed [$964,000 last month](https://x.com/lium%5Fio/status/2097824624117473549?ref=tao.media) across 1,187 renters, a 36% month-over-month increase. 

The next day, the project overtook [Chutes](https://www.tao.media/tag/chutes/) (notably just [named a new CEO](https://www.tao.media/chutes-names-christian-hyland-ceo-as-ai-compute-platform-expands/)) for the top subnet position by market capitalization following the [burning of $1 million worth of its alpha token](https://x.com/fish%5Fdatura/status/2097814414447481250?s=20&ref=tao.media) using product revenue. 

The asset's value has since increaed 10%.

![](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/2026/09/image-14.png)

Lium's update also included 1,908 new signups, 7,697 rentals, and roughly 80% month-over-month user retention. The team said 63% of rentals were started by software or agents, while the median rental lasted 1.15 hours.

Lium now draws capacity from [68 datacenters in 21 countries](https://x.com/lium%5Fio/status/2097803045362966828?s=20&ref=tao.media).

![](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/2026/09/image-13.png)

Earlier this month, Lium announced it had successfully ran Qwen3.6 35B at 14,499 output tokens per second on a rented NVIDIA B200 topping the model's LocalMaxxing board. 

The team put the implied cost at $0.107 per million output tokens, compared with $0.90 on the cheapest OpenRouter FP8 route it cited.

[Lium Tops LocalMaxxing Qwen3.6 35B Board With 14,499 Tokens Per SecondThe Bittensor SN51 team says the benchmark shows how rented GPU nodes can serve open models at lower per-token costs than hosted API providers.![](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/icon/Group-1321319358-36b83e59-79b4-4a3e-a303-50da8671c957.png)Intelligence | AI, Robotics and Bittensor News & InsightsTristan Hillerich![](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/thumbnail/ChatGPT-Image-Sep-8--2026--12_10_35-PM-c383f847-058b-43f0-b21c-1b405db8e8c3.png)](https://www.tao.media/lium-tops-localmaxxing-qwen3-6-35b-board-with-14-499-tokens-per-second/?ref=tao.media)