4x RTX PRO 6000
Full GLM-5.3
EXL3Recipe 1.1.0TensorFold v0.6.6
- Prose, single stream
- 92 tok/s
- Prose, 5 streams
- 140 tok/s
- Structured output, 5 streams
- 195 tok/s
- Prefill, up to
- 2,081 tok/s
- Up to 5 concurrent requests
- 1M context
Our mission
Open Access to Super Intelligence and the hardware that runs it
We research persistent agents and publish open source recipes for serving GLM-5.3 on RTX PRO 6000 GPUs.
Memory, identity & model orchestration
Open source · Apache-2.0
Model recipes.
Serve GLM-5.3 on RTX PRO 6000 GPUs with TensorFold.
./start.shbuilds, downloads and serves.
4x RTX PRO 6000
GLM-5.3-Flash
EXL3Recipe 1.2.0TensorFold v0.6.6
- Prose, single stream
- 266 tok/s
- Prose, 16 streams
- 1,087 tok/s
- Structured output, 16 streams
- 2,054 tok/s
- Prefill, up to
- 9,135 tok/s
- Up to 40 concurrent requests
- 1M context
2x RTX PRO 6000
GLM-5.3-Flash
EXL3Recipe 1.1.0TensorFold v0.6.6
- Prose, single stream
- 238 tok/s
- Prose, 4 streams
- 449 tok/s
- Structured output, 4 streams
- 651 tok/s
- Prefill, up to
- 6,908 tok/s
- Up to 4 concurrent requests
- 512K context
Measured with sparkDash, greedy, at 250 W per GPU. TensorFold by Ash Hart. Built in collaboration with Mia's AI Lab (EXL3 checkpoints and base patches).
What we do
Research & infrastructure.
Research & incubation
Research lab
Research toward Proto-AGI through memory, autonomy and model orchestration. We also incubate AI projects.
Task and workflow benchmarks
Model benchmarks
We test memory, tool use and reasoning in Hermes, with and without Protagine.
Launching in 2027
HPC investment SPVs
Special purpose vehicles for investment in HPC infrastructure.
Infrastructure access
We help infrastructure investors gain priority access to compute infrastructure allocations.
From the lab
Projects.
By request · Ongoing research
Proto AGI Engine
A harness for memory, autonomy and coordinated model roles. Built for Hermes.
In development
Our research into voice, vision, embodiment and out-of-model-weights recursive learning.
View A.E.V.A.Health technology
HealthForge
We incubated the AI research behind HealthForge, a personal health app.
Visit HealthForgeSelected work
NVIDIA B300 deployments.
We have supported several deployments across Central America and the United States.
About Aevonix Research
The team, our workand writing.