Open Superintelligence Factory.
- Tweets 3,087
- Following 49
- Followers 85,629
- Likes 44,145
Pinned Tweet
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
79
109
1,209
184,174
Prime Intellect retweeted
🙌 Great work from the team at @PrimeIntellect launching Prime Inference with GLM-5.3 served by vLLM at scale! 🚀
Getting agent workloads right takes more than fast decode. Their work covers prefill/decode topology, scheduler bubbles, and reliable tool calls.
Check it out!
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
8
11
157
13,129
Prime Intellect retweeted
you can basically do anything with:
1. human experts carefully curating high-quality domain specific Evals & Environments
2. a cracked infra provider that helps you (& your agents) train models
including “thermodynamic ML research” 👀
Using Prime Intellect, Extropic post-trained Qwen3.6-35B-A3B for thermodynamic ML research, nearly tripling its eval results on held-out tasks in ~100 GRPO steps.
They built a custom RL environment with verifiers and trained on Hosted Training, Prime Sandboxes, and Prime Inference. This meant Extropic didn't have to manage multi-node GPU infrastructure, so the team could focus on research.
5
74
8,104
Prime Intellect retweeted
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
79
109
1,209
184,174
Prime Intellect retweeted
we're organizing a small event at our office next wednesday during COLM, if your loss curve looks like this, join us :)
luma.com/colm-primeintellect…
10
9
235
15,833
RL needs more long-horizon tasks beyond math and coding. @niloofar_mire's team at CMU built an environment for drug design.
SMDD-Bench comprises 502 small-molecule design tasks with RDKit, ADMET-AI and Boltz-2 in the loop. The challenges go beyond chemistry: long-horizon planning, exploration, and learning from imperfect feedback are also open problems for RL/ML!
SMDD-Bench is available in our Environments Hub, ready to train with prime-rl. Thank you for sharing with the community!
One of the pivots I did over a year ago when I started as Faculty @LTIatCMU and @CMUEngineering was towards AI for Molecule Design and Drug Discovery, and today we are releasing a blog post + our @PrimeIntellect and @harborframework environments for our benchmark SMDD!
We find that: 1) Harness optimization can go a super long way in scientific tasks, and we are still heavily bottlenecked by the model capabilities rather than the knowledge, 2) Automating the harness optimization can work on some cases and not others!
Amazing work led by @SureshRaghu07, w/ @KevinH1119568 and @aviral_kumar2
12
19
198
15,009
Prime Intellect retweeted
I recently joined Prime, and in my first two weeks the sheer velocity and level of execution I've witnessed here are orders of magnitude higher than what I previously thought possible.
Today we're launching Prime Inference, and I got a front-row seat 😎
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
13
7
111
10,779
Prime Intellect retweeted
Introducing Prime Inference
Own your intelligence
Own your inference
Win
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
16
19
366
17,944
Prime Intellect retweeted
Prime Inference processes close to a trillion tokens a day across RL rollouts, synthetic data, long-horizon agents, and coding tasks. Now we’re bringing our open-source inference stack to the public—a key piece of Prime’s mission to build open intelligence for continual learning agent loop.
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
6
7
74
5,699
Prime Intellect retweeted
🦋 🫶 open source (like @vllm_project and @NVIDIAAI)
we will continue to hill climb @michellechen bench and contribute fixes upstream so everyone can benefit
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
4
9
119
6,654
Introducing Prime Inference:
We've served trillions of tokens for RL and dedicated customer deployments
To own your intelligence, you need to own your inference
Unpacking our inference stack
79
109
1,209
184,174
How to get started
prime inference chat 'z-ai/glm-5.3' "Write a haiku about KV caches.”
Or point any OpenAI SDK at api.pinference.ai/api/v1
2
1
52
3,731
Prime Inference is available today with serverless endpoints and reserved capacity.
Read the full technical breakdown:
primeintellect.ai/blog/prime…
4
7
95
4,432