Claude Code vs gh-aw
Compare 45 filters with recorded values on both sides. 23 of those values differ. How we select pairs.
| Open any value for its source | Claude Code | gh-aw | Add an ADE to this comparison | Empty column |
|---|---|---|---|---|
| Screenshot |  |  | ||
| Where you work | ||||
| Mac app | Yes | Yes | ||
| Windows app | Yes | Yes | ||
| Linux app | Yes | Yes | ||
| Browser | Yes | No | ||
| Native iOS | Yes | No | ||
| Native Android | Yes | No | ||
| Phone browser | Yes | No | ||
| Command line | Yes | Yes | ||
| Agents | ||||
| Claude Code | Yes | Yes | ||
| Codex | No | Yes | ||
| Cursor Agent | No | Yes | ||
| Gemini CLI | No | Yes | ||
| OpenCode | No | Yes | ||
| Pi | No | Yes | ||
| Grok | No | Not listed | ||
| Aider | No | Yes | ||
| Antigravity | No | Yes | ||
| Uses my subscription | Yes | Yes | ||
| Where it runs | ||||
| On your computer | Yes | Yes | ||
| On your own server | Yes | Yes | ||
| In the vendor's cloud | Yes | Yes | ||
| Keeps going with laptop closed | Yes | Yes | ||
| How it works | ||||
| Worktree per task | Yes | No | ||
| Containers | Yes | Yes | ||
| VM or microVM | Yes | Yes | ||
| Parallel agents | Yes | Yes | ||
| Coordinator agent | Yes | Yes | ||
| Built-in issues | Yes | Yes | ||
| Scheduled runs | Yes | Yes | ||
| Team and shipping | ||||
| Team sessions | Not listed | No | ||
| Team roles | Yes | No | ||
| SSO | Yes | No | ||
| Pull requests | Yes | Yes | ||
| Review screen | Yes | No | ||
| Ownership | ||||
| Product license | Proprietary | MIT | ||
| Self-hosting | Yes | Not listed | ||
| Telemetry opt-out | Yes | Yes | ||
| Price | ||||
| Has a free plan | No | Yes | ||
| Pricing model | Subscription, per seat, usage-based, bring your own key | Entirely free, bring your own key | ||
| Paid plans from | $11–20/month | No paid plan | ||
| Status | ||||
| Lifecycle status | Not listed | Not listed | ||
| Measured by ADEMark Third-party, 15 ADEs, to 12 November 2026 · ademark.dev, CC BY 4.0 · How to read places | ||||
| Cold start | 2.43 s · 7th of 12 | Not benchmarked | ||
| New thread, ready to type | Not measured | Not benchmarked | ||
| First word on screen | Not measured | Not benchmarked | ||
| Typing lag, 5 agents streaming | Not measured | Not benchmarked | ||
| Open a 2,000-message thread | Not measured | Not benchmarked | ||
| Memory at rest, 5 threads | 1.66 GB · 10th of 12 | Not benchmarked | ||
| Idle CPU | 0.71% · 4th of 13 | Not benchmarked | ||
| Memory under load | Not measured | Not benchmarked | ||
| Tokens to answer “hi” | 75.5K tokens · 12th of 13 | Not benchmarked | ||
| Tokens for one real question | 294.3K tokens · 12th of 13 | Not benchmarked | ||
| Team build: hidden tests passed | Not measured | Not benchmarked | ||
No recorded answers differ across these products.