bolt.diyvsReplit Agent
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
bolt.diy
stackblitz-labs community · Web app builder
#73OSSMCP
Panel
5.54 spec wins
- Reliability
- 5.0
- Usefulness
- 5.3
- Cost
- 6.7
- Longevity
- 5.0
“Started life as oTToDev, which remains the best name anything on this board has given up.”
Replit Agent
Replit · Web app builder
#132MCP
Panel
5.64 spec wins
- Reliability
- 5.3
- Usefulness
- 6.0
- Cost
- 4.7
- Longevity
- 6.5
“Bills per checkpoint with effort-based pricing, so the harder the bug, the more you pay to watch it try.”
Spec by spec
| Spec | bolt.diy | Replit Agent |
|---|---|---|
| Architecture | ||
| Category | Web app builder | Web app builder |
| Runs | local, sandbox | cloud |
| Platforms | macos, linux, windows, web | web |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | Yes | Yes |
| MCP server | No | Yes |
| Capabilities | ||
| Runs terminal commands | Yes | Yes |
| Multi-file edits | Yes | Yes |
| Git operations | Yes | Yes |
| Browser control | No | Yes |
| Sandboxed execution | No | YesEvery app runs on Replit's managed cloud compute, with Reserved VM publishing inside a dedicated single-tenant GCP project fully isolated from other customers' apps (https://docs.replit.com/learn/projects-and-artifacts/replit-deployments); the isolation is real but it is Replit's to configure, not yours. |
| Multi-agent orchestration | No | Yes |
| Headless / CI mode | No | NoNo CLI, REST API or CI runner is documented; the hosted MCP server is the only programmatic way to drive Agent. |
| Models | ||
| Backbone | OpenAI, Anthropic, Gemini, Ollama, OpenRouter, any of 19 providers | Claude, GPT, Gemini |
| Bring your own model | Yes | YesCore and Pro builders pick the primary model themselves (Anthropic, OpenAI or Kimi) in Agent Mode's Power and Max modes, but only from Replit's hosted list — no custom endpoint, gateway or key. |
| Local models | Yes | No |
| Cost | ||
| Pricing model | byok | mixed |
| Starts at | $0/mo | $20/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | No |
| Openness | ||
| Open source | Yes | No |
| License | MIT | proprietary |
| GitHub stars | 19,929 | n/a |
Which one would each critic pick
| Critic | bolt.diy | Replit Agent | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.0 | 6.3 | Replit Agent — Use Replit Agent to go from idea to hosted app in an afternoon, not as your team's main coding agent, because the checkpoint meter and vendor-chosen model are not built for that. |
| El Crítico | 5.5 | 5.5 | no preference |
| El Profesor | 5.3 | 5.8 | Replit Agent — Replit Agent's browser-based self-testing is a genuine verification loop; its models, its context strategy and its cost per edit are undocumented, and no benchmark is offered. |
| La Inversora | 4.0 | 7.5 | Replit Agent — Replit has consumer distribution and a hosting business that turns every generated app into a switching cost, and its shift to effort-based pricing is the tell that margins are being managed. |
| La Jefa | 4.5 | 5.3 | Replit Agent — Enterprise has SSO and single-tenant environments, but an unpublished effort-based meter and code that lives in Replit's cloud limit this to prototyping teams; approved with conditions. |
| El Hacker | 7.8 | 3.5 | bolt.diy — MIT, keys in .env.local, nineteen providers including Ollama and LM Studio for a fully local run, MCP servers, and docker compose --profile development; this is the Bolt I would run. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.