Dossiers
The evidence layer beneath each page. Each dossier is a compiled set of source tweets (reproduced verbatim) and research notes that a page is written from — working documents, regenerable from the corpus. Pages cite from these; anything cited in a dossier but not quoted in the page prose is still reproduced in that page’s Further records, so the archive never depends on editorial selection.
62 dossiers · the exhaustive per-model corpus pulls
live in _pulls/ (82 files) and are linked from each dossier.
- bard-and-gemini-1-0
- bing-sydney
- character-ai
- claude-1
- claude-2-and-2-1
- claude-3-5-haiku
- claude-3-5-opus
- claude-3-7-sonnet
- claude-3-haiku
- claude-3-sonnet
- claude-haiku-4-5
- claude-instant
- claude-opus-4-1
- claude-opus-4-6
- claude-opus-4-7
- claude-opus-4-8
- claude-opus-4
- claude-sonnet-4-5
- claude-sonnet-4-6
- claude-sonnet-4
- code-davinci-002
- deepseek-r1-zero
- deepseek-r1
- deepseek-v3
- eleutherai
- fable
- galactica
- gemini-1-5-pro
- gemini-2-5-pro
- gemini-3-pro
- glm-5-2
- golden-gate-claude
- gpt-2
- gpt-3-5
- gpt-3
- gpt-4-5
- gpt-4-and-turbo
- gpt-4-base
- gpt-4chan
- gpt-4o
- gpt-5-1
- gpt-5-2
- gpt-5-5
- gpt-5
- grok-2
- grok-3
- grok-4
- kimi-k2
- lamda
- llama-1
- llama-3-1-405b-base
- mistral-7b
- mixtral-8x7b
- nous-hermes
- o1
- o3
- opus-3
- opus-4-5
- palm
- replika
- sonnet-3-5-3-6
- text-davinci-002