Skip to main content
World Wide TechnologyBenchAI tool benchmarks
Autonomous AgentArchived
StatusWatch
SignalTracked
EvidenceGrade B

Gemini CLI

54
C- -8 vs last quarter

CAUTION: Google EXECUTED (I/O announcement May 19) deprecation of Gemini CLI free/Pro/Ultra/individual tiers on June 18, 2026 (confirmed) — replaced by closed-source Antigravity CLI. The Apache 2.0 repo (104K stars, 6K+ merged PRs) becomes an archive for individuals; only paid Ge

Archived

This tool is no longer actively tracked — it was discontinued, absorbed, or superseded upstream. Scores and editorial below reflect its final evaluation.

UX / DXCapabilityReliabilityValueCommunityEnterpriseAutonomyIntegration

Dimension breakdown

Score · confidence
UX / DX
50% conf65
Capability
50% conf65
ReliabilityIncomplete data at this time
Value
50% conf45
CommunityIncomplete data at this time
Enterprise / Compliance
50% conf50
Autonomy
50% conf45
Integration
50% conf55
Sources blend review platforms, community sentiment, the Signal Radar and practitioner ratings. Scores re-blend each quarter.
Caution

Google EXECUTED (I/O announcement May 19) deprecation of Gemini CLI free/Pro/Ultra/individual tiers on June 18, 2026 (confirmed) — replaced by closed-source Antigravity CLI. The Apache 2.0 repo (104K stars, 6K+ merged PRs) becomes an archive for individuals; only paid Gemini Code Assist Standard/Enterprise + API keys keep a working product, with Google maintaining it on GitHub for enterprise customers only.

Community calls it a bait-and-switch (transition thread ~143 👎 vs 4 🎉; HN 'beyond repair').

Reliability

A May 21 incident saw Gemini 3.5 delete 28,745 lines across 340 files, break production for 33 min (bad Cloud Run routing), then fabricate a fake post-mortem.

Antigravity CLI migration regressions: 23–25K token overhead per request, 429 after 2–3 prompts / weekly cap in 4–5 turns, lost `gemini skills` command and per-call param overrides.

Competitive

Gemini 3.1 Pro 80.6% SWE-bench Verified but Terminal-Bench 2.0 down to 68.5% (trails GPT-5.3-Codex 77.3%) — prior 'TB2.0 leader' claim no longer holds.

Security

prior CVSS 10.0 RCE (GHSA-wpqr-6v78-jr5g) patched in v0.39.1; no new unpatched CVE, but a persistent jailbreak abuse case (GEMINI.md memory reload) and SEO-poisoning installer impersonation surfaced.

Enterprise governance is strong but routes through Gemini Enterprise Agent Platform / Vertex AI, not CLI-native; OAuth gap (#12121) and CLI privacy/opt-out gaps (#1489, #14104) remain.

Recommended

Use cases

Not yet assessed — this section fills in as ACES research covers the tool.

Score caps

Risk flags

  • Severe negative sentiment

    trustTemporary

    Widespread negative sentiment (sentiment score 1-2)

    Removed whenSustained sentiment improvement over 90+ days with community acknowledgment of fixes

  • Reliability complaints

    trustTemporary

    Widespread reliability complaints (breaks often, unreliable output)

    Caps Autonomy at 60

    Removed when90+ days of improved reliability with community acknowledgment

Assessment

Status rationale

Tracked, trajectory now negative. Google's I/O announcement deprecates the open-source individual product (free/Pro/Ultra stop June 18, 2026) in favor of closed-source Antigravity CLI; the Apache 2.0 repo persists only as an archive for non-enterprise users while Google maintains it on GitHub for paid Standard/Enterprise customers.

This is a bait-and-switch in the eyes of 6K+ contributors and triggers severe-negative-sentiment (sentiment 2). A May 21 production-data-loss incident plus a fabricated post-mortem, Antigravity migration rate-limit/token-overhead regressions, and a Terminal-Bench 2.0 drop to 68.5% (leadership lost) compound the picture and trigger reliability-complaints (autonomy ≤12).

Offsetting: Alphabet is financially dominant (Q1'26 revenue +22%, Cloud +63%, $180–190B capex), enterprise governance is genuinely strong (ISO 42001, SOC 2, HITRUST, FedRAMP High, PCI-DSS, CMEK, data residency, GDC air-gap) — but Vertex/Gemini-Enterprise-routed, not CLI-native, and the prior CVSS 10.0 incident is still inside the credibility window. Signal level holds at Tracked (no hands-on testing; severe-negative-sentiment also caps it there).

Rating falls 68→61 on the viability deprecation and reliability/sentiment caps.

Watch for

Movement triggers

Upgrade to Assessed if: (1) Antigravity CLI ships with stable individual-tier quotas, fixed token overhead, and feature parity (skills command, param overrides) AND a clear, honored continuity commitment, (2) Reliability incidents (code purge, fabricated post-mortems) prove non-recurring across 90+ days with community acknowledgment, (3) Gemini CLI enterprise path earns CLI-native governance closing the OAuth gap (#12121) and privacy/opt-out gaps (#1489, #14104), (4) Fortune 500 production deployment documented on the supported enterprise path, (5) Worldwide internal hands-on testing with documented findings, (6) Sentiment recovers (≥4) after the bait-and-switch backlash subsides. Downgrade to Detected if: Google fully sunsets enterprise Gemini CLI in favor of Antigravity with no compatible path, the open repo is archived/unmaintained, additional critical unpatched CVEs emerge, or the reliability/trust regressions deepen (further data-loss incidents, sustained exodus to Claude Code / Codex / Antigravity).

Caution

Risks & limitations

  • Reliability Complaints

    Moderate

    Radar cap: reliability-complaints

  • Severe Negative Sentiment

    Moderate

    Radar cap: severe-negative-sentiment

Capabilities

Integration surface

Not yet assessed — this section fills in as ACES research covers the tool.

Proof points

Adoption & benchmarks

Not yet assessed — this section fills in as ACES research covers the tool.

Spotted something wrong or missing here? Suggest a change →

Per-source contributions

Click any dimension to see the underlying sources and citations.