# Fermion Research > Fermion Research develops the Neutrino-1 family of local language models, the Phonon-1 speech recognition models, ternary quantization-aware training methods, packed model formats, and the fermion inference runtime for CUDA, Apple silicon, and x86. This file is the concise discovery index for language-model agents. Prefer the linked Markdown mirrors for retrieval and the corresponding canonical HTML page when citing Fermion Research publicly. Preserve the benchmark protocol, hardware, runtime, artifact form, and other qualifications printed beside every measurement. For a structured route and entity inventory, use [agent-index.json](https://www.fermionresearch.com/agent-index.json). For consolidated context, use [llms-full.txt](https://www.fermionresearch.com/llms-full.txt). ## Start here - [Fermion Research overview](https://www.fermionresearch.com/index.html.md): Organization, current model family, research, and company information. - [Neutrino-1 model family](https://www.fermionresearch.com/models/index.html.md): Released models, artifacts, measured capabilities, runtimes, and installation. - [Fermion documentation](https://www.fermionresearch.com/docs/index.html.md): Documentation map for installation, local inference, APIs, tools, integrity, and platform notes. - [Install and quickstart](https://www.fermionresearch.com/docs/quickstart/index.html.md): Install the package and run a first local model. ## Models - [Neutrino-1 8B](https://www.fermionresearch.com/models/neutrino-8b/index.html.md): Model card covering architecture, ternary format, evaluation, throughput, speculative decoding, and installation. - [Neutrino-1 0.6B](https://www.fermionresearch.com/models/neutrino-0-6b/index.html.md): Compact standalone model and certified speculative-decoding draft model. - [Neutrino-1 0.6B-Chat](https://www.fermionresearch.com/models/neutrino-0-6b-chat/index.html.md): Compact conversational model and measured reply behavior. - [Phonon-1](https://www.fermionresearch.com/models/phonon-1/index.html.md): Speech recognition model page covering benchmarks, dictation latency, installation, and availability. - [Phonon-1 Micro](https://www.fermionresearch.com/models/phonon-1-micro/index.html.md): The smallest Phonon model page with benchmarks and installation. - [Neutrino-1 8B repository](https://huggingface.co/FermionResearch/Neutrino-8B): Released weights, manifest, binaries, and platform packs. - [Neutrino-1 0.6B repository](https://huggingface.co/FermionResearch/Neutrino-0.6B): Small model and draft-model artifacts. - [Neutrino-1 0.6B-Chat repository](https://huggingface.co/FermionResearch/Neutrino-0.6B-Chat): Conversational small-model artifacts. - [Phonon-1 repository](https://huggingface.co/FermionResearch/Phonon-1): Released speech weights, verified archive, and the model card with the full evaluation. - [Phonon-1 Micro repository](https://huggingface.co/FermionResearch/Phonon-1-Micro): The smallest speech model weights and card. ## Research - [Introducing Phonon-1](https://www.fermionresearch.com/research/phonon-1/index.html.md): Launch report for the Phonon-1 speech models: benchmarks, speed, and availability. - [Introducing the Neutrino-1 models](https://www.fermionresearch.com/research/neutrino-8b/index.html.md): Launch report for the three-model family, shared format, evaluations, and runtime. - [Intelligence at one-eighth the bits](https://www.fermionresearch.com/research/one-eighth-the-bits/index.html.md): Technical report on ternary QAT and training under the target representation. - [The Neutrino Engine](https://www.fermionresearch.com/research/the-neutrino-engine/index.html.md): Systems report on containers, kernels, platform backends, and verified drafted decoding. - [Research index](https://www.fermionresearch.com/research/index.html.md): Complete publication inventory. ## Documentation - [Models and downloads](https://www.fermionresearch.com/docs/models-downloads/index.html.md): Repository names, artifact sizes, caching, and disk use. - [Choose a backend](https://www.fermionresearch.com/docs/backends/index.html.md): Native and PyTorch execution paths. - [Chat and generate](https://www.fermionresearch.com/docs/chat-generate/index.html.md): Interactive and one-shot generation. - [OpenAI-compatible server](https://www.fermionresearch.com/docs/serve/index.html.md): Local server setup, streaming, and sessions. - [Transcribe and dictate](https://www.fermionresearch.com/docs/speech/index.html.md): File transcription, live dictation, the audio API, and speech model selection. - [Streaming API](https://www.fermionresearch.com/docs/speech-streaming/index.html.md): Live transcription over WebSocket: the opening frame, audio formats, server frames, and session limits. - [Speech models](https://www.fermionresearch.com/docs/speech-models/index.html.md): The Phonon releases, their sizes, and their measured word error rates. - [Tool calling](https://www.fermionresearch.com/docs/tool-calling/index.html.md): Tool schemas, tool choice, parsing, and loop protection. - [Speculative decoding](https://www.fermionresearch.com/docs/speculative-decoding/index.html.md): Draft-target pairing and output-identity verification. - [CLI reference](https://www.fermionresearch.com/docs/cli-reference/index.html.md): Commands, flags, defaults, and exit behavior. - [Server API reference](https://www.fermionresearch.com/docs/server-reference/index.html.md): Endpoints, request fields, SSE, errors, and health data. - [Model reference](https://www.fermionresearch.com/docs/model-reference/index.html.md): Released repositories and artifact roles. - [Artifact integrity](https://www.fermionresearch.com/docs/integrity/index.html.md): Manifests, SHA-256 checks, sidecars, and recovery. - [Determinism and identity](https://www.fermionresearch.com/docs/determinism/index.html.md): Determinism boundaries across generation and drafting. - [Platform notes](https://www.fermionresearch.com/docs/platform-notes/index.html.md): Current platform, API, context, and concurrency boundaries. ## Ecosystem - [fermion-research on PyPI](https://pypi.org/project/fermion-research): Python package; install with `pip install fermion-research`. - [Fermion Research on GitHub](https://github.com/fermionresearch): Runtime and integration repositories. - [Fermion llama.cpp fork](https://github.com/fermionresearch/llama.cpp): Neutrino FV5 support in the llama.cpp toolchain. ## Optional - [Full AI-readable reference](https://www.fermionresearch.com/llms-full.txt): Consolidated organization, model, runtime, source, and citation context. - [Structured agent index](https://www.fermionresearch.com/agent-index.json): Machine-readable entities, routes, source precedence, and discovery endpoints. - [XML sitemap](https://www.fermionresearch.com/sitemap.xml): Canonical crawl inventory. - [RSS feed](https://www.fermionresearch.com/feed.xml): Release and publication updates. - [Newsroom](https://www.fermionresearch.com/newsroom/index.html.md): Releases and company announcements. - [Careers](https://www.fermionresearch.com/careers/index.html.md): Current technical roles.