Models Powering the Grove — Phoenix Grove Systems
Phoenix Grove · Model Intelligence

Models Powering the Grove

The AI landscape has changed. Open-source models now meet and exceed the intelligence of billion-dollar proprietary systems. Phoenix Grove is built on that shift.

For years, frontier AI intelligence was locked behind proprietary walls. If you wanted the best models, you used the platforms that built them, on their terms, under their privacy policies, with their limitations.

That era is ending. A wave of open-source and open-weight models from labs around the world has closed the intelligence gap to single digits. Models that rival the output quality of the most expensive closed-source systems are now available, deployable, and combinable in ways that weren't possible even twelve months ago.

Phoenix Grove exists because of this shift. We don't spend billions training foundation models from scratch. We select the most capable open-source models available, run them on privacy-first US-based infrastructure with zero data retention at the inference level, and orchestrate them into cognitive systems that think in ways no single model can.

The fleet

Some of the models we work with

Our model fleet is constantly evolving. As the open-source landscape advances, so do we. These are some of the models currently powering Phoenix Grove's cognitive architectures.

Kimi K2.6 (Coming soon in Snowlion Update)
Moonshot AI · Open Weights · 1T Parameters

A trillion-parameter mixture-of-experts model leading open-source agentic and reasoning benchmarks. Exceptional at sustained, multi-step problem solving and tool use across complex workflows.

MiMo 2.5 Series
Xiaomi · 1T Parameters

A frontier-class model that quietly topped usage charts across major API providers before anyone knew who built it. Strong general intelligence with particularly impressive reasoning depth at remarkably efficient cost.

GLM-5
Zhipu AI (Z.AI) · Open Weights

The leading open-source model on SWE-bench coding evaluations, with top-tier performance across knowledge, reasoning, and real-world software engineering tasks.

Gemma 4 31B
Google DeepMind · Open Weights · Apache 2.0

Google's open-weight flagship, punching far above its parameter class. Multimodal support, efficient token usage, and the permissive licensing that makes true deployment independence possible.

This is a partial view. Our fleet includes additional models across different capability tiers, and we evaluate new releases continuously. Phoenix Grove AI releases new cognitive builds and model updates on a regular basis.

Intelligence

Independently verified

The models in our fleet consistently meet, match, or come within striking distance of the leading closed-source models on comprehensive independent intelligence testing, including evaluations across reasoning, coding, agentic tool use, knowledge accuracy, and instruction following.

We don't ask you to take our word for it. Artificial Analysis maintains one of the most rigorous independent benchmarking platforms in the industry, evaluating hundreds of models across standardized conditions. You can look up every model in our fleet and compare them directly against any proprietary model from any lab.

The gap between open-source and closed-source AI is no longer a generation. It's single digits on a standardized index. And it's closing every month.

Benchmark performance varies by task domain and evaluation methodology. We encourage independent verification. Scores reflect individual model capabilities; Phoenix Grove's cognitive architecture combines multiple models, which is not captured by single-model benchmarks.

Beyond individual models

We build cognitive systems.

Phoenix Grove doesn't point you at a model and step back. Every build in our lineup is a cognitive system: models working together, blending inputs and outputs across multiple layers to produce responses with a depth and dimensionality that no single model achieves on its own.

Some of that orchestration is visible to you in our multi-core thinking panels. Some of it runs beneath the surface: models coordinating on search, memory retrieval, safety, context synthesis, and reasoning in ways that shape every response you receive.

The models are the ingredients. The cognitive architecture is the recipe. And the recipe is what makes Phoenix Grove feel different from talking to a model directly.

Privacy-first deployment

Open-source models, privately deployed

Many of the most powerful open-source models originate from labs outside the US. They run on privacy-first, US-based infrastructure so your data never has to leave.

US-based infrastructure. All model inference runs on US-based servers with zero data retention at the inference level. Your conversations are processed domestically regardless of where the model weights were originally published.
Your data never touches the original labs. Open-source means the model weights are publicly available. The models powering Phoenix Grove run on privacy-first infrastructure entirely separate from their creators. Your prompts and conversations never route through the original developers' servers.
No training. No telemetry. No exceptions. We don't train on your conversations. We don't collect behavioral telemetry. Your conversation data is stored securely in our own database, protected by our privacy architecture.

See what the Grove can do.

Frontier intelligence. Privacy-first architecture. A cognitive system that thinks with you.

Try Phoenix Grove AI
Currently in paid beta · Adults only · Cancel anytime