summarize

July 6, 2026 · View on GitHub

summarize [input] [flags]

Takes a URL, local file path, or - for stdin and produces a summary. With --extract, prints cleaned content instead of summarizing; stdin is not supported in extract mode.

Synopsis

summarize "https://example.com/article"
summarize ./report.pdf --model openai/gpt-5-mini
summarize "https://youtu.be/..." --slides --slides-ocr
pbpaste | summarize -
summarize "https://example.com" --extract --format md > clean.md
summarize "https://example.com" --json --metrics detailed

Input

  • URL — http/https. YouTube, podcast hosts, generic articles, raw media.
  • File path — PDF, image, audio, video, plain text, Markdown, HTML.
  • - — read content from stdin (text or binary).

If [input] is omitted, summarize prints concise help and exits.

Flags

Extraction

--extract : Print extracted content and exit. No LLM call. Supports URLs, YouTube/direct media, local audio/video, and local PDFs; - stdin is not supported. The --extract-only alias is hidden but still works. : Extraction can still use configured remote transcription, OCR, or Markdown providers for media/PDF inputs.

--format <format> : md or text. Controls website extraction format and whether files are preprocessed to Markdown for model compatibility. Default: text. Default in --extract mode for URLs: md.

--firecrawl <mode> : off, auto, always. auto falls back to Firecrawl when local extraction looks blocked or thin. Needs FIRECRAWL_API_KEY.

--preprocess <mode> : off, auto, always. Default auto. Controls markitdown usage for files. always only affects file inputs. : Image-only PDFs can retry through OpenAI vision OCR when OPENAI_API_KEY is set. Set MARKITDOWN_OCR_MODEL to choose the OCR model, or MARKITDOWN_OCR_DPI to tune rendered page size.

--markdown-mode <mode> : off, auto, llm, readability. For websites: HTML → Markdown strategy. For YouTube/transcripts: llm formats raw transcripts into clean Markdown with headings.

--max-extract-characters <count> : Cap printed characters in --extract mode. Default: unlimited.

YouTube / video

--youtube <mode> : auto, web, apify, yt-dlp, no-auto (skip auto-generated captions). Default auto.

--transcriber <name> : Select the local transcription stage: auto, whisper, parakeet, canary. Default auto — Groq still runs first when keyed, then local ONNX/whisper.cpp, then cloud fallbacks.

--video-mode <mode> : auto, transcript, understand. understand prefers vision/video understanding when supported.

--timestamps : Include timestamps in transcripts when available.

--diarize [provider] : Add speaker labels to YouTube, local audio/video, or direct media URLs. Provider: auto, elevenlabs, or openai.

--identify-speakers / --no-identify-speakers : Resolve generic diarization labels to names, or disable configured resolution for one run. Uses authoritative anchors and OpenAI GPT-5.5 context inference.

--speaker-profile <name> : Select a reusable profile from speakers.profiles in the config file.

--speaker-at <timestamp=name> : Name the speaker active at a timestamp. Repeat the flag for multiple speakers.

--remember-speakers : Atomically persist resolved names, anchors, and a transcript-hash-guarded source mapping under the selected profile.

Slides

--slides [value] / --no-slides : Extract slides from a YouTube URL, direct video URL, or local video file and render inline alongside the summary. Combine with --extract to interleave slides in the full transcript. See Slides mode. : Use --no-slides to disable slide extraction enabled in the config file for one run.

--slides-ocr / --no-slides-ocr : Run OCR on extracted slides. Requires tesseract. : Use --no-slides-ocr to disable configured slide OCR while retaining slide extraction.

--slides-debug : Print slide image paths instead of rendering inline (useful in non-image terminals).

--slides-dir <dir> : Output directory base. Default ./slides.

--slides-scene-threshold <value> : Scene detection threshold, 0.1–1.0. Default 0.3.

--slides-max <count> : Cap on extracted slides. Default 6.

--slides-min-duration <seconds> : Minimum gap between slides. Default 2.

Summary control

--length <length> : short, medium, long, xl, xxl (or s/m/l) or a character cap like 20000 or 20k. Default long. Override globally via output.length in Config.

--force-summary : Always run the LLM, even when extracted content is shorter than the requested length.

--language, --lang <language> : auto, en, de, english, german, … Default auto (matches source). Configurable via output.language.

--prompt <text> / --prompt-file <path> : Override the summary prompt. Content is appended after the override.

Models

--model <model> : Model id. auto, <config-preset>, cli/<provider>/<model>, xai/..., openai/..., nvidia/..., minimax/..., google/..., anthropic/..., zai/..., github-copilot/..., or openrouter/<author>/<slug>. Default auto. See LLM overview.

--cli [provider] : Use a logged-in CLI provider: claude, gemini, codex, agent, openclaw, opencode, copilot, agy, pi. Equivalent to --model cli/<provider>. Without a value: enable CLI providers in auto-selection.

--thinking <effort> : OpenAI reasoning effort. none, low, medium, high, xhigh. Aliases: off, min, mid. Only affects reasoning-capable OpenAI models.

--service-tier <tier> : OpenAI service tier. default, fast, priority, flex. Maps to service_tier on the request.

--fast : Shortcut for OpenAI service_tier=priority. Same effect as --service-tier fast.

--max-output-tokens <count> : Hard cap for LLM output tokens. Accepts 2000, 2k. Overrides provider defaults.

--retries <count> : LLM retry attempts after timeouts or transient API failures. Default 1.

Output

--json : Structured JSON envelope on stdout. Includes prompt + metrics. Disables streaming.

--stream <mode> : auto, on, off. Default auto (TTY-only). Disabled in --json mode.

--plain : Keep raw text/Markdown — no ANSI/OSC rendering.

--no-color : Disable ANSI colors.

--width <columns> : Override terminal width for Markdown rendering. Default: auto-detect, max 120.

--theme <name> : Pick a CLI theme. Run summarize --help for the list. Env: SUMMARIZE_THEME.

Cache

--no-cache : Bypass the summary (LLM) cache. Media + transcript caches stay enabled.

--no-media-cache : Disable the media download cache (yt-dlp outputs).

--cache-stats : Print cache stats and exit.

--clear-cache : Delete the cache database and exit.

See Cache internals for what's cached and where.

Diagnostics

--verbose : Detailed progress on stderr.

--debug : Alias for --verbose. Also defaults --metrics to detailed.

--metrics <mode> : off, on, detailed. Default on. Adds a compact second-line breakdown on stderr when detailed.

--timeout <duration> : Cap the whole pipeline. 30, 30s, 2m, 5000ms. Default 2m.

-V, --version : Print version and exit.

Examples

# Extracted plain text from a URL.
summarize "https://example.com" --extract

# Extracted Markdown — prefers Firecrawl when configured.
summarize "https://example.com" --extract --format md

# YouTube transcript as cleanly-formatted Markdown.
summarize "https://www.youtube.com/watch?v=..." \
  --extract --format md --markdown-mode llm

# ElevenLabs diarization with verified speaker names.
summarize "https://www.youtube.com/watch?v=..." --extract \
  --diarize elevenlabs --identify-speakers \
  --speaker-profile my-podcast --speaker-at "0:12=Host Name" --remember-speakers

# Inline slides + summary.
summarize "https://www.youtube.com/watch?v=..." --slides

# Full transcript + inline slides.
summarize "https://www.youtube.com/watch?v=..." --slides --extract

# Hard caps + a specific model.
summarize "https://example.com" \
  --length 20k --max-output-tokens 2k --timeout 2m \
  --model openai/gpt-5-mini

# OpenAI fast tier with reasoning.
summarize "https://example.com" \
  --model openai/gpt-5.5 --fast --thinking medium

# GitHub Models via GITHUB_TOKEN.
summarize "https://example.com" --model github-copilot/gpt-5.4

# Config preset (alias defined in ~/.summarize/config.json).
summarize "https://example.com" --model mymodel

# Strict JSON for scripts.
summarize "https://example.com" --json --verbose

# Summarize the clipboard.
pbpaste | summarize -

Environment

The provider keys and binary paths picked up at runtime are listed in LLM overview. SUMMARIZE_MODEL overrides the default model selection; SUMMARIZE_THEME picks the theme.

See also