Claude Code vs mini-SWE-agent
Claude Code (Anthropic) and mini-SWE-agent (Princeton University & Stanford University (SWE-agent team)) side by side on public data, as of Nov 20, 2026.
Claude Code has 6.1x the weekly package downloads of mini-SWE-agent (18.9M against 3.1M). Claude Code has 16.0x the GitHub stars of mini-SWE-agent (156K against 9,771). Over the last 30 days downloads moved +16% for Claude Code and -2% for mini-SWE-agent. Claude Code shipped 72 stable releases in the past 90 days.
Which should you choose?
Choose Claude Code if
- you want it inside your editor
- you want to work from a browser
- you want to check in from your phone
- you want subagents or several agents working in parallel
- you prefer the more widely used tool, with more docs and answers around (Gid Score 88 against 51)
- you want frequent updates (72 stable releases in 90 days against 0)
Choose mini-SWE-agent if
- you want to start free: it has a free plan, Claude Code does not
- you want code you can read, fork or self-host (MIT)
- you need to run local models
- benchmark results matter to you: 91.0% on Terminal-Bench against 65.8%
Built from documented features, official prices, status pages and usage data on this page; it does not judge output quality.
Side-by-side numbers
| Metric | Claude Code | mini-SWE-agent |
|---|---|---|
| Gid Score | 88 | 51 |
| Usage data as of | Nov 19 | Nov 19 |
| Rank by Gid Score | #1 | #40 |
| Downloads, last 7 days | 18.9M | 3.1M |
| Downloads, 7-day change | up 15.1% | down 3.5% |
| Downloads, 30-day change | up 15.5% | down 1.7% |
| VS Code installs | 30.6M | — |
| Marketplace rating | 3.7 / 5 (936) | — |
| GitHub stars | 156K | 9,771 |
| Homebrew installs, 30 days | 61.7K | — |
| Hacker News mentions, 30 days | 1,889 | 5 |
| Latest stable release | 2.1.328 · yesterday | 2.4.6 · 4mo ago |
| Stable releases, 90 days | 72 | 0 |
| Terminal-Bench (best run) | 65.8% · v4.1 | 91.0% · v4.1 |
| Price | $20/mo | Free (OSS) |
| Usage limits | Pro and Max: a 5-hour session limit plus a weekly limit; Ultra: a weekly limit only (verified Ultra accounts have no 5-hour limit) Pro: More usage than Free every 5 hours and every week ($17/month with yearly billing); the plans page gives no message counts all plans | — |
| Incidents, last 30 days | 15 · 11 major or critical | — |
| License | Proprietary | MIT |
| First released | February 2025 | April 2024 |
| Interfaces | CLI, IDE extension, Desktop, Web, Mobile, SDK | CLI |
| Models | Anthropic Claude models via Claude plans, a Console API key or supported third-party cloud providers | Any model via LiteLLM, OpenRouter, Portkey and similar |
A dash means there is no public data for that metric as of Nov 20, 2026.
Trends
Claude Code
Anthropic's agentic coding tool that reads a codebase, edits files and runs commands from the terminal, with VS Code and JetBrains integrations plus desktop, web and mobile access. It runs Claude models through a Claude subscription, an Anthropic Console API key or supported third-party cloud providers.
Pricing: Claude Code is included in Pro ($17/month annual or $20/month monthly), Max ($100/month), and Ultra ($200/month) plans. The Free plan does not include Claude Code. Team plans offer standard seats at $20/seat/month annual ($25 monthly) and premium seats at $100/seat/month annual ($125 monthly), both with Claude Code. Enterprise seats cost $20/seat/month annual (20 seat minimum) plus usage at API rates.
Latest release 2.1.328:
- Added auto mode for Claude Opus 4.6 and Sonnet 4.6 on Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry and Claude apps gateway; interactive sessions with no permission mode start in it
- Added "onFailure": "block" for mcp_tool hooks, as for command and HTTP hooks: a hook whose call fails blocks the action
- Added a warning when a message from a session that skips permission prompts may be held and dropped by a cloud session
- Added a notice in /status and claude doctor when an OpenTelemetry exporter has no endpoint and its data would likely be lost
mini-SWE-agent
Open-source software-engineering agents from Princeton and Stanford researchers: SWE-agent (2024) and its roughly 100-line successor mini-SWE-agent, which uses only bash and works with any model through LiteLLM and similar routers. The SWE-agent README says development effort has moved to mini-SWE-agent.
Pricing: Free MIT-licensed research software; users pay their model provider through their own API keys.
Latest release 2.4.6:
- fix(agents): count billed calls that fail to parse against cost_limit
Claude Code vs mini-SWE-agent: common questions
How do Claude Code and mini-SWE-agent prices compare?
Claude Code: $20/mo. mini-SWE-agent: Free (OSS). Prices come from each vendor's pricing page.
Is Claude Code or mini-SWE-agent open source?
mini-SWE-agent is open source (MIT); Claude Code is proprietary.
Which is more popular, Claude Code or mini-SWE-agent?
Claude Code has a Gid Score of 88 against 51 for mini-SWE-agent. The score combines package downloads, editor installs, GitHub activity and attention, updated daily.
Do Claude Code and mini-SWE-agent support local models?
mini-SWE-agent documents support for local models; Claude Code's documentation does not.
Do Claude Code and mini-SWE-agent support MCP servers?
Claude Code documents support for MCP servers; mini-SWE-agent's documentation does not.