Subnets / SN14
Cacheon SN14
Cacheon (subnet 14) is a Bittensor subnet held by 2,197 unique coldkeys. Its liquidity pool holds 25.7k τ, with the alpha token trading at 0.0093 τ. Roughly 7% of tracked holders are in profit — most bought in above today's price. Seller exhaustion is elevated: 77% of recent sellers have emptied their wallets, thinning the supply overhang. Full insider flow, whale tracking and the complete proprietary factor set for Cacheon are available to SubnetStats subscribers.
ALPHA PRICE
0.0093 τ
IN TAO
HOLDERS
2,197
UNIQUE COLDKEYS
POOL DEPTH
25.7k τ
TAO IN POOL
7D MOMENTUM
-1.5%
ALPHA vs TAO
Project activity
Cacheon’s DeepSeek arena is heating up again.
The challenge has been rebuilt to test performance under a more demanding, production-like inference workload, yet miners are still finding sizeable gains.
Latest winner:
→ +10.88% evaluated workload speedup
→ 179.2 → 191.5 tok/s de
Meta released Muse Spark Open on 15 October.
Six days later, it's live in Cacheon's optimization arena.
Can miners find the kind of inference efficiency gains they've found on DeepSeek V4.1-Flash, GLM-5.3, and Qwen 3B?
Let's find out. https://t.co/wqExP29qJk
DeepSeek has open-sourced DeepInfer, its optimized inference engine.
With performance benchmarks showing great margins for inference providers of close to 80%.
Cacheon, however, remains ahead of the curve.
Our latest DeepSeek-V4.1-Flash benchmark: 434 tokens/s per user, servin
Need faster Inference? https://t.co/SIRXnRCudG https://t.co/GVBG0kO1on
We're excited to be providing inference on the Liquid Inference exchange.
Token volume is picking up, so this presents a great opportunity to monetize our optimized catalog of models.
Prices are currently competitive but we expect to be out-competing and delivering on revenue.
More throughput. More ways to compete.
A 50% throughput gain on the same hardware, at equivalent quality, opens up new opportunities for teams building and serving AI:
→ Lower costs: offer more competitive prices, or make your own model cheaper to run and use
→ Compete on capac
Why DeepSeek-V4.1-Flash for our next competition?
Because targeting a frontier open model keeps Cacheon at the leading edge.
Strong coding and agent performance. Native visual understanding. An architecture designed for efficiency.
How much faster can we run DeepSeek-V4.1-Flas
A little summary of our performance progress so far on #SN14.
Speeds are measured using AIPerf which simulates multi-user concurrent workloads.
Can you do better? Submit your improved kernel and earn SN14-Alpha on Bittensor $TAO
Need a speed-boost? Reach out and we'll tailor t
Affine's King model. Two weeks of optimization. 40.9% more throughput.
@affine_io became Cacheon’s first customer. These are the performance numbers behind that partnership.
We pushed Qwen3.6-35B to 528.9 tokens/s, up from 375.4 tokens/s on stock SGLang.
We’re building on this
$100k in revenue last month. @affine_io became our first customer.
We’re now working to monetize our Qwen and GLM-5.3 optimizations through licensing deals and our own inference service.
This quarter, we’re targeting more customers and expanding competitions to high-volume mode
Q3 is over. How are we doing against our roadmap?
The proof of concept is built:
✓ Multiple models optimized - MiniMax-M3., GLM-5.3 and Qwen3.6.
✓ Competitions run in parallel across models.
✓ Speed improvements evaluated end to end, with output-quality checks.
Different model
AI inference is getting cheaper. But it’s consuming more and more of the AI compute stack.
McKinsey released a new report yesterday, estimating that by 2030, inference will account for ~60% of AI compute demand, versus ~40% for training.
That makes inference optimization increa
Great to see the work we’ve been doing with @affine_io get a moment on stage at @exploitsummit.
Here’s @const_reborn telling the story in his opening keynote ↓ https://t.co/E6f1zfxtXa
Qwen3.6 arena is now open for submissions!
Better yet, we've generalized the evaluation system to support arbitrary kernel changes.
No more waiting on slot updates to submit your winning kernel!
We're excited to see what you can do. Submit today and earn Alpha right away.
GLM-5.3 hits a new milestone on Cacheon: ~12.5% above its earlier benchmark, measured under different test conditions.
On SN14 in $TAO Bittensor, miners optimize GPU operations to accelerate inference. We test changes with the full model running, and accepted work can support fu
In our last competition, Cacheon miners pushed MiniMax M3 throughput to 2,337.9 tok/s: +32.2% vs the SGLang baseline.
What might that kind of performance improvement be worth commercially?
+32.2% throughput = ~61% more profit.
How? SemiAnalysis InferenceX™ models a B300 runnin
Happy Labour Day!
We've been hard at work over the weekend.
Announcing our GLM-5.3 arena!
Miners can now submit kernel optimizations and get rewarded by the protocol.
GLM-5.3 presents a huge opportunity for production-use and monetization of the subnet.
More info: https://t.
Another great result coming out of this competition's arena.
We've hit 2,274 tok/s on our MiniMax-M3 baseline
We're hard at work on integrating the next top model for our arena.
Cacheon focuses on production-ready models for real enterprise use-cases with evaluations only via
And there we have it.
A measured speedup of over 12% on MiniMax-M3. Hitting over 1950 tok/s
This submission validates the thesis of the subnet and we're excited for more to roll-in.
In the meantime we're integrating top open-weights models to optimize for production use. https
We've lowered the submission fee to 0.5 TAO
We hope this change will increase submissions from new miners and amp-up the competition.
Think you can do better?
Submit your kernel at https://t.co/SIRXnRCudG and get rewarded today
Or follow along at https://t.co/hnpmOEFkWI
Cacheon is now open for submissions!
Check out our docs to get started mining today: https://t.co/9umGZa8Tcn
Attention Cacheon stans!
The subnet code is public and ready for launch.
We're targeting Monday at 9am EDT.
We're eager to get feedback on the mining process and our docs.
Please check the repo out below. We'll address anything that needs to be done for launch so we can star
https://t.co/IR52jgZmam
We've just locked 10k Alpha to show our conviction
Cacheon continues.
We're excited to show you what's next. We've brought on a new LLM engineer and are dedicating more resources to bring you the top inference optimization competition on Bittensor, supported by real expertise.
Cacheon has parted ways with @xavi3rlu.
We're excited for the future of Cacheon and will keep you updated on the next phase of the subnet.
SELLERS DRAINED
77%
28D SELLERS EMPTIED OUT
CAPITULATION
-0.57
BOUNCE COMPOSITE (z)
HEAT 7D
0.14×
VOLUME ÷ POOL DEPTH
HOLDERS IN PROFIT
7%
TRACKED WALLETS · ENTRY BELOW PRICE
WHAT'S CARRYING THE PRICE · 30DORGANICemission rank #27 · above the bar (rank 32)
30D RETURN
+5.8%
ALPHA vs TAO
WITHOUT CHAIN BUYS
+1.9%
SAME 30 DAYS, PROTOCOL BID REMOVED
CHAIN SHARE OF BUYING
60%
CHAIN BUYS vs TRADERS' NET
CHAIN BUYS / DAY
17 τ
PROTOCOL BUYING ITS ALPHA
See who's buying and selling Cacheon — every trade classified miner / validator / owner / outsider, plus holder cost basis and the full 23-factor screener.
OPEN THE INSIDER TAPE →