Skip to main content
World Wide TechnologyBenchAI tool benchmarks
Autonomous Agent
StatusIn Review
SignalAssessed
EvidenceGrade B

Devin

73
B -1 vs last quarter

Fully autonomous AI software engineer operating from web, Slack/Teams, CLI/Terminal, Windsurf IDE, and REST API v3.

UX / DXCapabilityReliabilityValueCommunityEnterpriseAutonomyIntegration

Dimension breakdown

Score · confidence
UX / DX
79% conf51
Capability
79% conf50
ReliabilityIncomplete data at this time
Value
50% conf95
CommunityIncomplete data at this time
Enterprise / Compliance
50% conf80
Autonomy
50% conf80
Integration
50% conf80
Sources blend review platforms, community sentiment, the Signal Radar and practitioner ratings. Scores re-blend each quarter.

Fully autonomous AI software engineer operating from web, Slack/Teams, CLI/Terminal, Windsurf IDE, and REST API v3. Devin 2.2 (Feb 24) self-review + auto-fix; through May 2026 ships Devin-Manages-Devins parallel orchestration, Review auto-fix/auto-merge, PR resuming, GitLab interactive PR review, and Adaptive model routing (SWE-1.6 default, opus/sonnet/codex selectable).

Windsurf acquisition thesis now realized — Devin runs embedded in the IDE, not a separate tab. Series D CLOSED May 27: $1B at $25B pre / $26B post (2.5x in 8 months), ~$492M ARR run-rate, enterprise usage +50% MoM for 6 months; Devin now writes 89% of Cognition's own code.

Enterprise base deepened: Mercedes-Benz (8-month legacy modernization in 8 days), NASA, Goldman Sachs, Citi, Santander, Itaú (70% of security vulns auto-fixed), Infosys multi-vertical. SOC 2 Type II + ISO 27001:2022 + CCPA; VPC dedicated deployment, training opt-in by default.

SWE-bench 45.8% now mid-field as Cursor (65.7%) and frontier agents reach 80%+. Under active evaluation for backlog/migration offload. Monitor for: hands-on WWT validation, unaddressed prompt-injection attack surface (Critic + egress allowlist are partial mitigations), and ACU cost unpredictability.

Recommended

Use cases

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

Score caps

Risk flags

No active caps — no risk flags apply to this tool right now.

Assessment

Status rationale

Assessed because strong enterprise deployment evidence (Infosys multi-vertical, Goldman Sachs, Citi, Mercedes-Benz, NASA, Santander, Nubank, Mercari, Itaú), major capability advances (Devin 2.2 self-review/auto-fix, Devin-Manages-Devins orchestration, Adaptive model routing, Windsurf embedded), and extraordinary company momentum (Series D closed May 27 — $1B at $26B post-money, ~$492M ARR, +50% MoM for six months). Blockers for Validated: hands-on internal evaluation not completed; prompt-injection attack surface unaddressed (Critic + default-deny egress allowlist are partial mitigations, no published CVE/patch closing it); mixed community sentiment (Trustpilot 3.0/5, score 3, need 4+); ACU pricing unpredictability (no pre-task cost disclosure).

Watch for

Movement triggers

Upgrade to Validated if: hands-on internal testing completed at ≥40% complex task success on representative WWT workloads; prompt injection vulnerability addressed with published patch/CVE disclosure; community sentiment improves to 4+; ACU billing adds pre-task cost estimation (addresses predictability concern); and explicit human approval granted. Downgrade to Tracked if: $25B funding round fails with significant valuation reset; prompt injection produces a named enterprise security incident; or community sentiment drops to 2 or below.

Caution

Risks & limitations

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

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.