Spent some time looking for setups. Nothing really popped out except the developing M15 compression on TAO. Think I'll have to wait for some more pa to develop. Might be H1 200 EMA retests.
Bittensor
1D
TODAY
7D
WTD
30D
MTD
90D
QTD
12M
YTD
Mindshare
Displays the percentage of social media posts mentioning each token on this blockchain. Higher percentages indicate greater social attention and discussion around that particular cryptocurrency.
GAINS
ABSOLUTE
STACKED
AREA
Hippius offers storage at a lower price per terabyte than Google Drive, iCloud, and Dropbox.
On the latest Novelty Search, @Hippius_cloud breaks down how SN75 turns Bittensor miners into S3-compatible cloud storage, with no egress fees for moving la...
8/21/2026 10:49
17K
8
21
121
3
Meta's Muse Glimmer 30B is live on Chutes at $0.0245 per million input tokens, $0.0978 out. That ties the cheapest tier we serve. Images and text in, 131K context.
It runs inside a hardware-attested TEE with end-to-end encryption, like every model w...
8/20/2026 13:30
3.5K
12
15
88
2
Novelty Search :: Subnet 75 :: @Hippius_cloud
Hippius is building decentralized cloud storage on Bittensor, turning distributed hardware into S3-compatible infrastructure.
Miners provide storage and serve data across the network, earning for the ca...
8/19/2026 08:12
21K
8
18
127
5
Qwen shipped Qwen3.8-27B's weights on Friday. Dense 27B, Apache 2.0, and the model half of X spent the weekend benchmarking. It is already live on Chutes.
Qwen3.8-27B on Chutes: 262K context, text + image input, $0.40 per million input tokens, $3.00...
8/17/2026 19:00
4.8K
10
19
104
7
.@404gen_ just launched Atlas, its AI workflow platform for 3D production.
They already work with multiple AAA game studios like Square Enix. And Parralel's Colony goes live this month with real-time asset generation powered by Atlas.
How Bittensor...
8/13/2026 16:35
26K
16
29
160
4
Send a prompt to a hosted AI API and it arrives on someone else's machine as readable text. Staff. Logs. The routing layer. The company that owns the model.
Nothing in the architecture stops any of them from reading it. What stops them is a policy....
8/13/2026 13:30
2.3K
2
12
76
2
New on Chutes: docuextract.
Feed it a PDF or an image. It hands back clean markdown or schema-guided JSON.
Invoices, contracts, bank statements, scanned forms, research papers. Multi-page documents in a single call.
And like 100 percent of the mod...
8/12/2026 13:30
6.3K
9
20
111
10
This week on Novelty Search :: Bittensor Subnet 17, @404gen_
404GEN is building open models for generative 3D. Its latest release, Titan v1, turns images into production-ready 3D assets, with open weights and end-to-end generation in around 30 secon...
8/11/2026 20:42
22K
8
28
183
7
Someone hands your training run an update. How do you know they did the work you asked for, and not something else?
Jon Durbin's answer in Parallax is that the acceptance test is concrete. Did this update improve loss on the routed expert it was sup...
8/10/2026 18:00
2.7K
3
9
49
3
Chutes is migrating from our current smart contract mechanism to conviction lock.
Conviction lock is a better version of our smart contract, now enforced at the protocol level rather than by a contract we deploy.
The network has grown, and the mecha...
8/10/2026 16:46
17K
9
27
174
7
4.7M parameters per expert where a standard expert carries 12.6M. Same shape, but roughly one third of the size.
That's a LatentMoE expert, and the trick is almost shocking in its simplicity. A standard expert runs 2048 wide in and out. The LatentMo...
8/10/2026 13:30
3.1K
2
18
98
2
Liquid commodity subnets just autoscale supply and revenue with higher emissions. All of it available to other mechanisms, cause now you can simply “mine the compute you need to train with”. Anything I mine just default swaps for more compute/inferen...
8/8/2026 21:52
18K
9
31
221
3
Recently, 5 more enterprise platforms worth $22B+ combined have been testing @_redteam_’s technology. They reach 900M+ accounts and handle 1B+ weekly transactions.
RedTeam is set to unveil a new immune system built to anticipate cyber threats.
Here...
8/8/2026 12:40
102K
17
60
380
35
“We’re already in a state of permissionlessness far beyond any example of a similar incentive network.”
On the latest Novelty Search, Bittensor co-founder @const_reborn looks ahead at the network’s roadmap.
With Root Reborn now live, the next focus...
8/7/2026 19:55
16K
9
53
267
11
Every eval table compares scores and price. Ours has one more row: whether anyone could see your tokens at inference time.
As published by each lab: Kimi K3 scores 88.3 on Terminal-Bench 2.1 and 93.5 on GPQA Diamond. GLM-5.2 scores 81.0 on Terminal-...
8/7/2026 18:00
3.4K
5
15
96
3
Two models. One gets more answers right and costs a fortune. The other gets slightly fewer and costs almost nothing.
Jon Durbin's question is which one you should actually want. His answer is that solve rate on its own is a vanity number, because it...
8/7/2026 16:30
3.1K
4
18
76
2
A power company had to choose between its residents and a new data center. It picked the data center.
Jon Durbin on how fragile centralized AI actually is: a stray bullet through a fiber line in Texas took out a data center. Ship anchors rip backbon...
8/6/2026 18:00
3.8K
6
17
77
6
Train a model across machines on the open internet and the first question is always the same one. What do those machines see of my data?
For Parallax the answer is: nothing readable. Workers never see raw training tokens or raw activations. What a w...
8/6/2026 16:00
3.2K
5
18
84
4
Targon Organizations are officially live. Team collaboration on https://t.co/buFR01d46X now enables:
→ Shared workloads with one bill
→ Owner, admin, and member roles
→ Org-scoped tokens for CI/CD
→ Confidential compute, now for teams
Invite your t...
8/5/2026 20:16
8.1K
8
21
88
5
This Thursday on Novelty Search :: Root Reborn & Ecosystem Updates
Root Reborn is now live on mainnet.
Join us to unpack one of the network’s largest upgrades for Root stakers to date and explore how Root Reborn changes validator incentives and...
8/5/2026 17:01
15K
20
39
233
5
Introducing Root Reborn: a new market for optimizing TAO yield across Bittensor.
Validators compete to build the best-performing subnet baskets. Successful strategies can attract more stake, so more root yield goes toward the subnets that create the...
8/4/2026 13:40
76K
35
120
554
66
Hippius alphanomics: the next stage is live.
We laid this out in our original alphanomics roadmap, and we're now delivering it. It's a complex system, so we've built it deliberately and rolled it out in careful stages. Here's what just went on-chain...
8/4/2026 13:36
3.8K
3
10
37
4
Recurrent attention would not sync. Jon Durbin tried the existing decentralized algorithms, the whole DiLoco family included, and none of them translate.
So he built a new sync method into Parallax, exact and fully non-blocking, within 0.6% of the c...
8/3/2026 19:00
4.2K
5
21
98
10
.@theminos_ai, the team behind Bittensor SN107, co-authored a new field report with @OpenAI on agentic scientific computing.
It features HelixForge, the GPU-native engine powering the subnet’s open competition to improve mutation detection. https://...
8/3/2026 17:30
11K
11
28
154
3
165,335 tokens per second. That is Parallax pre-training a 5B test model on eight consumer RTX 5090s.
Optimized DDP training on eight RTX Pro 6000s managed 147,708. Hardware that costs roughly 3x more.
Parallax won even on the cheaper hardware. htt...
8/3/2026 13:30
3.7K
6
29
128
3
"We decided to push Quasar to 120 billion parameters.”
On the latest Novelty Search, @QuasarModels unveiled its new 120B-parameter foundation model, pretrained on 7.5T tokens.
After early training on SN24 and further work on the team’s 120-H200 clu...
7/31/2026 18:39
9K
7
41
177
9
Train a big AI model across machines on the open internet and you hit a wall. Every single step, the machines have to ship hundreds of gigabytes of expert traffic to each other. The next step cannot start until all of it lands.
Parallax removed that...
7/31/2026 13:30
4.7K
4
24
110
5
DFLASH cuts decode time by 30 to 50 percent on tested models.
A small model guesses the next few tokens. The big model checks the guess and keeps
what it got right. If the guess is wrong, the big model writes the token itself.
gemma-4 31B and qwen3...
7/30/2026 13:30
2.2K
6
13
62
0
Image boss has been beaten!
Another improvement brought by Gradients miners to image models training...
Here is what they did to get on the throne:
- Dynamic checkpoint selection, they hold out 3–6 images, train past the usual step count with frequ...
7/29/2026 11:30
1.2K
4
4
22
2
Moonshot shipped Kimi K3's weights yesterday. 2.8T parameters, the largest open-weight model released to date. It is already live on Chutes.
Kimi K3 on Chutes: 1M context, native text + image + video input, $3.00 per million input tokens, $15.00 out...
7/28/2026 16:31
5.3K
16
28
153
5
This Thursday on Novelty Search :: SN24 Quasar
@QuasarModels is building decentralized foundation model training on Bittensor :: creating an open marketplace for training long-context AI models.
SN24’s incentive mechanism :: Miners compete by train...
7/28/2026 14:10
20K
15
41
187
9
.@jon_durbin pre-trained a 20B MoE for under $10 an hour of compute.
DropZone Episode 1 is him explaining how, in full.
- a surrogate 48x smaller than the real expert, standing in for it
- workers that train on data they never see
- what ternary we...
7/28/2026 14:04
10K
7
32
131
21
.@bitmind just published a paper placing BitMind Forensics among the world’s top deepfake detection systems.
It matched the best commercial image detector, beat the best commercial video detector, and outperformed top models on major benchmarks.
He...
7/27/2026 17:29
53K
6
49
215
12
Mindshare Trend
Mindshare statistics in social posts for tokens on this blockchain. '30D' represents total mindshare percentage over the past 30 days, '∆30D' shows the percentage change during this period, and 'TREND' indicates the current directional movement of social media attention.
Top Bullish Mindshare
Shows the 6 tokens on this blockchain with the highest number of positive mindshare in social media posts.
ALL
EX-MAJORS
Top Bearish Mindshare
Shows the 6 tokens on this blockchain with the highest number of negative mindshare in social media posts.
ALL
EX-MAJORS