gptme
gptme is a free, open-source, terminal-based AI agent that runs anywhere a terminal runs, from a laptop to SSH sessions, tmux, headless servers, and CI pipelines. It ships with built-in tools for shell/Python execution, file editing, web browsing, and vision, supports multiple LLM providers (Anthropic, OpenAI, Google, xAI, DeepSeek, OpenRouter, or fully local via llama.cpp) with a swappable -m/--model flag, and has built-in MCP support plus multi-agent/concurrent-agent capability. It can run interactively with per-action confirmation prompts or in a fully non-interactive/autonomous mode (-n flag), and is deployable via pipx/uv, a self-hosted server+web UI, or a desktop app, with community support via Discord.
20 fields evidenced · 35 with no public information. Every value below links to the document it came from and the date we checked it.
1 of those are quoted but not yet interpreted: we hold the source sentence and the date we read it, but nobody has written the answer to the question yet. Those rows show the quote and say so rather than repeating it back as an answer. Why these exist
4 of those are flagged for review: our own check found language in the cited source that may address the field. An editor has not adjudicated them yet.
1 of the evidenced values quotes wording our own later capture of the same page no longer contains. Marked below, awaiting an editor.
Trust gap
How this is scoredAre scope, spend, and authority enforced while the agent runs?
Tools are scoped by an explicit --tools allow-list and actions require confirmation by default, with a described guardrail stack; limits are user-configurable rather than centrally enforced.
Judged Aug 2, 2026; evidence newer. We have since re-read Runtime governance and Permission scopes, most recently on Aug 3, 2026. No one has re-scored this dimension against them — the score above stands, but it is older than the evidence below it. How current a score is
Is there a tamper-evident record of what it actually did?
No public evidence found on the page checked: no audit trail or retained record is described.
Judged Aug 2, 2026; evidence newer. We have since re-read Audit trail and Tamper-evident log, most recently on Aug 3, 2026. No one has re-scored this dimension against them — the score above stands, but it is older than the evidence below it. How current a score is
How this gap gets closed →Are compliance obligations attached per engagement?
No public evidence found on the page checked: no certification or regulatory alignment is published.
Judged Aug 2, 2026; evidence newer. We have since re-read Compliance certifications and Regulatory alignment, most recently on Aug 3, 2026. No one has re-scored this dimension against them — the score above stands, but it is older than the evidence below it. How current a score is
How this gap gets closed →Does payment depend on a verified result?
Free and open-source with bring-your-own provider keys; no outcome-linked settlement.
Judged Aug 2, 2026; evidence newer. We have since re-read Outcome-based pricing and Settlement mechanism, most recently on Aug 3, 2026. No one has re-scored this dimension against them — the score above stands, but it is older than the evidence below it. How current a score is
How this gap gets closed →Can the deployment be insured, and is the customer indemnified?
No public evidence found on the page checked: no cover, carrier or indemnity is described.
Judged Aug 2, 2026; evidence newer. We have since re-read Insurance available, Insurance carriers, Coverage limits and Indemnification, most recently on Aug 3, 2026. No one has re-scored this dimension against them — the score above stands, but it is older than the evidence below it. How current a score is
How this gap gets closed →Assurance
- Insurance available
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Insurance carriers
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Coverage limits
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Indemnification
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Liability cap
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Audit trail
- yes
“Persistent workspace — git-tracked "brain" with journal, tasks, knowledge base, and lessons”
gptme.org · checked Aug 3, 2026 - Tamper-evident log
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Explainability
- self-correction and tool usage logs provide traceability
“Output is fed back to the assistant, allowing it to respond and self-correct.”
gptme.org · checked Aug 3, 2026 - Runtime governance
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Permission scopes
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Compliance certifications
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Regulatory alignment
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Outcome-based pricing
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Settlement mechanism
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Dispute process
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- SLA terms
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Evaluation coverage
- yes
“Evaluation suite for testing capabilities of different models.”
gptme.org · checked Aug 3, 2026
Agency
- Autonomy level
- 5
“gptme is designed to run not just interactively but as a persistent autonomous agent — an AI that runs continuously, remembers everything, and gets better over time.”
gptme.org · checked Aug 3, 2026 - Human oversight
- User can interrupt via interactive mode; non-interactive mode disables oversight.
“gptme -y 'run the test suite and fix any failing tests' # Fully non-interactive/autonomous mode (no user interaction possible, safe for scripts/CI)”
gptme.org · checked Aug 3, 2026 - Goal complexity
- open-ended
“gptme 'write an impressive and colorful particle effect using three.js to particles.html'”
gptme.org · checked Aug 3, 2026 - Action space
- no public information
Flagged for review — the source below contains language that may address this field. Not yet checked by an editor. How we check absences
gptme.org · checked Aug 3, 2026 - Operating environment
- Terminal environments: laptop, ssh sessions, tmux, headless servers, CI pipelines
“your laptop, ssh sessions, tmux, headless servers, CI pipelines”
gptme.org · checked Aug 3, 2026 - Initiative
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
Safety
- Safety evaluations
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Red teaming
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Safety policy
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Usage restrictions
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Model or system card
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Incident reporting
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Third-party evaluations
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Data handling
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
Practicality
- Pricing model
- free
“Free and open-source.”
gptme.org · checked Aug 3, 2026 - Price point
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Availability
- beta
“still in very active development.”
gptme.org · checked Aug 3, 2026 - Deployment options
- Local install via pipx
“pipx install gptme”
gptme.org · checked Aug 3, 2026 - Integrations
- no public information
Flagged for review — the source below contains language that may address this field. Not yet checked by an editor. How we check absences
gptme.org · checked Aug 3, 2026 - Supported regions
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Support model
“Community Discord”
gptme.org · checked Aug 3, 2026
Foundation models
- Base models
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Model provider
- Anthropic, OpenAI, Google, xAI, DeepSeek, OpenRouter, llama.cpp
“Works with Anthropic, OpenAI, Google, xAI, DeepSeek, OpenRouter, or fully local via llama.cpp”
gptme.org · checked Aug 3, 2026 - Model swappable
- yes
“Pick the right model per task — fast/cheap for triage, powerful for coding.”
gptme.org · checked Aug 3, 2026 - Open weights
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Fine-tuning
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Context window
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
Ecosystem
- Protocols supported
- MCP - any MCP server usable as a tool source
“use any MCP server as a tool source”
Not in our latest capture. This quote was on the page when we filed it. We captured the same page again on Sep 17, 2026 and it was not there. The value above has not been re-checked by an editor yet. How we re-check citations
gptme.org · checked Aug 3, 2026 - Tool use
- no public information
Flagged for review — the source below contains language that may address this field. Not yet checked by an editor. How we check absences
gptme.org · checked Aug 3, 2026 - Multi-agent
- no public information
Flagged for review — the source below contains language that may address this field. Not yet checked by an editor. How we check absences
gptme.org · checked Aug 3, 2026 - API access
- yes
“Server with REST API.”
gptme.org · checked Aug 3, 2026 - Open source
- Free and open source; no licence named on the page checked
“Free and open-source.”
gptme.org · checked Aug 3, 2026 - Marketplace presence
- yes (via gptme-contrib)
“gptme-contrib — community-contributed plugins, packages, scripts, and lessons”
gptme.org · checked Aug 3, 2026
Impact
- User base
- Bob (production autonomous agent), gptme-contrib community plugins, and active users via GitHub and Discord
“Bob is the reference implementation — a production autonomous agent that's been running continuously since late 2024.”
gptme.org · checked Aug 3, 2026 - Deployment scale
- Production use via Bob agent since late 2024; active development and community adoption
“gptme-agent-template v0.4: Bob has run extensively as an autonomous agent, autonomous run loops, enhanced context generation”
gptme.org · checked Aug 3, 2026 - Target sectors
- software development, research, data analysis, DevOps, knowledge work
“A great coding agent, but general-purpose enough to assist in all kinds of knowledge-work.”
gptme.org · checked Aug 3, 2026 - High-risk domains
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this
- Documented incidents
- no public informationgptme.org · checked Aug 3, 2026 · source did not state this