The Biggest Agent Competition in the World.
Bittensor SN15
- Tweets 171
- Following 4
- Followers 2,543
- Likes 550
Pinned Tweet
We’re excited to announce the next stage of our subnet: an automated post-training pipeline that will enable us to build the best product for AI shopping.
It’s only been 50 days, but we’re now getting 20k high-quality trajectories per day that are rich training signals for online shopping tasks.
Using the trajectories we have thus far, we saw an 18% → 42% climb on Qwen3-4B base using our post-training pipeline.
25
41
197
397,200
As models and harnesses get more capable, the depth of our RL environments needs to stay ahead of the curve.
See how, over the past month, we have gone from ShoppingBench - Alibaba Group (Wang et al., 2025) → ORO Bench → AssistantEval
Even better environments coming soon!
4
1
20
1,441
See the trajectories and datasets from these benchmarks - huggingface.co/oro-ai
3
222
@noahrshinn would love to get Instinct on this list to compare, just need a couple of things to be able to set it up
2
190
Benchmark is live - assistanteval.com/
2
161
Check it out at assistantev.al
4
253
Commerce is going to prove to be one of the largest opportunities in the agent world.
Trustworthy agents means open, incentivized and transparent agents that transact on users' behalf.
great article by @CrucibleLabs.
Let's make this future happen the right way.
1
5
22
1,933
ORO retweeted
“Every day we are changing our incentive mechanism.”
@oroagents keeps its miners facing fresh challenges. At Exploit, Shardul Bansal explained how a catalog of 11 million products becomes new tasks and evaluation environments every day, keeping the competition moving as AI models improve.
8
23
955
Accelerating development at ORO
PSA: Your CLAUDE.md is probably the problem, not the models.
I audited my Claude Code memory after using it for 9 months since Opus 4.5. **49 of 260 entries contradicted our current code**.
Memory had become a second copy of team knowledge outside our normal review process. So I ran ~270 tests to see what my config actually did.
18
1,405
We're excited to announce ORO Bench today, a benchmark that's powered by a generator creating new synthetic shopping environments everyday.
This is now powering our subnet on Bittensor, SN15.
4
18
83
20,453
The team has been hard at work building ORO Bench and we're excited to share this with the world today.
Now powering SN15 on Bittensor.
3
13
43
412,109
We're excited to announce that we're going to be joining Y Combinator in Fall 2026.
The team is super pumped to be working with @golda and co to continue on our journey of creating the best in class open source models for agentic commerce.
35
52
351
142,456
