Model Observatory
Copyright 2026 alphaBell. Viewer code is licensed under MIT.

Model configuration references:
RecursiveSI/flint, Recursive Superintelligence, Inc.
https://huggingface.co/RecursiveSI/flint
Qwen/Qwen3.8-27B, Qwen Team
https://huggingface.co/Qwen/Qwen3.8-27B
The publishers distribute these models under Apache License 2.0:
https://www.apache.org/licenses/LICENSE-2.0

Configurations record architectural facts. The Qwen configuration describes
its text decoder; the source model also includes other components.
Bundled measured activation captures are published by alphaBell for reuse
with this viewer. No model weights are included.

Capture engine reference: llama.cpp, MIT License
https://github.com/ggml-org/llama.cpp
A source patch and experiment harness are included in experiments/reproduce.
The patch applies to the pinned public llama.cpp source and retains its MIT license.
No compiled inference executable or model weights are included in this Space.

Qwen measurements use alphaBell Qwen Open, a post-trained derivative.
Its separate checkpoint release uses alphaBell's research and noncommercial license.
Parent architecture configurations are distinct from captured checkpoint identity.