Frontier-level intelligence at every scale. From the uncompromising power of our flagship Parallax 2 to the extreme efficiency of the upcoming Parallax 2 Mini, we are bringing elite reasoning to the hardware you actually own.
Two distinct models. One shared philosophy: absolute privacy, zero subscriptions, and frontier capability running entirely offline.
Parallax 2
Flagship
Our ultimate local intelligence engine. Trained on over 202,000 elite traces, Parallax 2 is designed for complex, multi-step tasks. It features a massive 262K context window, allowing it to read entire codebases, process long research papers, and act as an autonomous agent. It is the pinnacle of local performance.
Extreme efficiency without the compromise. Parallax 2 Mini is a marvel of distillation. Designed to run on just a few gigabytes of RAM, it can operate on almost any device—from older laptops to edge devices. It brings the Parallax family to the masses without sacrificing the core reasoning capabilities.
Context Window262,144 Tokens
Target HardwareLaptops / Edge Devices
Use CaseFast, Local Chat
The Baseline
The Parallax Standard
Regardless of which model you choose, these core principles apply to the entire family.
100% Local & Private
Your data never leaves your device. No APIs, no cloud inference, no subscriptions. Just pure, local compute.
Elite Data Curation
Both models are trained on the same elite reasoning, coding, and agentic traces that make the Parallax family so capable.