Gladia CLI

The Gladia CLI (

gladia

) transcribes pre-recorded audio from the terminal. Local files are auto-uploaded; URLs are passed directly to the API.

CLI vs SDK: use the CLI for quick terminal workflows when
gladia
is on PATH. For app integration or audio intelligence not exposed as CLI flags, see gladia-sdk-integration and gladia-audio-intelligence.

When to Use

User mentions gladia-cli,
```
gladia transcribe
```
, or wants quick terminal transcription
Agent should transcribe a local file or URL, then answer follow-up questions about the content
One-off transcription without writing application code

Prerequisites: verify

gladia --version

succeeds and an API key is configured (

GLADIA_API_KEY

~/.gladia

, or

--gladia-key

When NOT to Use

Live / real-time audio — use gladia-live-transcription
Building an app or CI pipeline in code — use gladia-sdk-integration
Audio intelligence not in CLI (translation, summarization, NER, PII, audio-to-LLM) — use gladia-audio-intelligence with the SDK

References

Consult these resources as needed:

./references/cli-vs-sdk.md -- CLI vs SDK routing: which features have CLI flags and which require SDK skills
../gladia-pre-recorded-transcription/SKILL.md -- SDK pre-recorded workflow and options
../gladia-audio-intelligence/SKILL.md -- Addons beyond CLI flags
../gladia-troubleshooting/SKILL.md -- API key, upload, and polling errors
gladia-cli repository

Setup

Install (macOS/Linux):

bash

curl -fsSL https://github.com/gladiaio/gladia-cli/releases/latest/download/install.sh | sh

Auth — get a key at app.gladia.io/account. Three options:

bash

# 1. Environment variable (preferred for CI and shells)
export GLADIA_API_KEY=<API_KEY>

# 2. Persist locally
gladia auth set <API_KEY>

# 3. Pass per command (global flag, works on any command)
gladia transcribe meeting.wav --gladia-key <API_KEY>

Credential order:

GLADIA_API_KEY

→

~/.gladia

→

--gladia-key

(first match wins)

List valid language codes:

gladia languages

Commands

Command	Description
`gladia transcribe <source>`	Transcribe a local file or `http(s)` URL
`gladia auth set <key>`	Save API key to `~/.gladia`
`gladia languages`	List supported ISO 639-1 codes

Transcribe Flags

Flag	Default	Description
`-o` , `--output`	`text`	`text` , `json` , `json-full` , `srt` , `vtt`
`--language`	—	Expected language(s), comma-separated ( `en` or `en,fr,de` )
`--code-switching` , `--code-switch`	off	Detect language per utterance
`--diarize`	off	Speaker identification
`--model`	API default	`solaria-1` or `solaria-3`
`-v` , `--verbose`	off	Show progress while polling

Global:

--gladia-key

— API key override

Agent Workflow: Transcribe Then Q&A

Check CLI —
```
gladia --version
```
; install or use SDK skills if missing
Pick flags — match output format and options to the user's question (tables below)
Run —
```
gladia transcribe <source> [flags]
```
; capture stdout
Answer — ground responses only in captured output; cite speakers and timestamps when available
Re-run if needed — if the question requires data not in the current output (e.g. timestamps, speakers), re-transcribe with different flags
Long audio — use
```
-v
```
for progress; for very long transcripts, redirect stdout to a temp file and read selectively

Do not invent transcript content. If output is empty or unclear, say so and suggest different flags or SDK skills.

Output Format Selection

User need	CLI approach
Plain transcript	default or `-o text`
Who spoke when	`--diarize -o text` or `-o json`
Timestamps per utterance	`-o json` (utterance list with `time_begin` , `time_end` )
Full API payload	`-o json-full`
Subtitle file	`-o srt` or `-o vtt` (add `--diarize` for speaker labels)
Model choice	`--model solaria-1` or `--model solaria-3`

Language behavior

You can list all the possible languages compatible with gladia with the command

gladia languages

Goal	Command
Auto-detect	`gladia transcribe <source>`
Constrain detection	`--language en,fr,de` (does not enable code switching)
Code switching	`--code-switching` (+ optional `--language` hints)

CLI vs SDK (summary)

For full CLI vs SDK routing, see ./references/cli-vs-sdk.md.

Feature	CLI	If not in CLI
Basic transcription	`gladia transcribe`	—
Speaker diarization	`--diarize`	gladia-audio-intelligence for advanced config
Language / code-switch	`--language` , `--code-switching`	SDK for advanced `language_config`
Translation, NER, PII, sentiment, audio-to-LLM	No	gladia-audio-intelligence
API summarization addon	No	SDK (agent may summarize `-o text` output informally)
Live streaming	No	gladia-live-transcription

Examples

bash

gladia transcribe meeting.wav
gladia transcribe https://example.com/podcast.mp3 -o json
gladia transcribe call.wav --diarize -o srt
gladia transcribe interview.mp3 --language en,fr --code-switching -v
gladia transcribe podcast.mp3 --model solaria-1 -o json-full
gladia transcribe meeting.wav --gladia-key <API_KEY>   # inline key, no env or ~/.gladia needed

Common Mistakes

Treating
--language en,fr
as code-switching — it only constrains detection; add
```
--code-switching
```
separately for per-utterance language detection
Answering without re-running — timestamp or speaker questions need
```
-o json
```
or
```
--diarize
```
; plain text may lack required fields
Inventing CLI flags — translation, summarization, NER, and PII have no CLI flags today; route to SDK skills
Using SDK code when CLI is requested — if the user has
```
gladia
```
installed and wants terminal workflow, run shell commands
Skipping auth check — transcription fails without a valid API key in env,
```
~/.gladia
```
, or
```
--gladia-key
```

For API errors and diagnostics, see gladia-troubleshooting.

gladia-using-cli

NPX Install

Tags

SKILL.md Content

Gladia CLI

When to Use

When NOT to Use

References

Setup

Commands

Transcribe Flags

Agent Workflow: Transcribe Then Q&A

Output Format Selection

Language behavior

CLI vs SDK (summary)

Examples

Common Mistakes

Further Reading