LettavsRaven
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
Letta
Letta · Agent harness
OSS
Panel
6.51 spec wins
- Reliability
- 6.0
- Usefulness
- 6.7
- Cost
- 7.0
- Longevity
- 6.3
“It used to be called MemGPT, and it remembers that.”
Raven
EverMind AI · Agent harness
OSS
Panel
6.26 spec wins
- Reliability
- 5.3
- Usefulness
- 6.3
- Cost
- 8.0
- Longevity
- 5.2
“It has a Sentinel that sends you proactive nudges, so the agent now has opinions about your afternoon.”
Spec by spec
| Spec | Letta | Raven |
|---|---|---|
| Architecture | ||
| Category | Agent harness | Agent harness |
| Runs | local, cloud | local, sandbox |
| Platforms | macos, linux, windows, web | macos, linux, windows |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | No | No |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | Yes | Yes |
| Multi-file edits | Yes | Yes |
| Git operations | No | No |
| Browser control | No | YesDeep Research performs web search and page reading via MiroThinker and returns a cited report; there is no general browser-automation tool documented. |
| Sandboxed execution | No | YesOnboarding asks for a sandbox or execution location, documented separately under docs/sandbox/usage.md. |
| Multi-agent orchestration | No | YesThe agent loop spawns subagents, and tracing records parent-child subagent runs. |
| Headless / CI mode | No | Yes |
| Models | ||
| Backbone | any | OpenAI, Anthropic, Gemini, OpenRouter, DeepSeek, MiniMax, Moonshot, Groq, Azure OpenAI, GitHub Copilot (OAuth), Codex (OAuth), Ollama, vLLM |
| Bring your own model | Yes | Yes |
| Local models | No | Yes |
| Cost | ||
| Pricing model | mixed | byok |
| Starts at | $20/mo | $0/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | Apache-2.0 | Apache-2.0 |
| GitHub stars | 25,000 | 5,061 |
Which one would each critic pick
| Critic | Letta | Raven | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.5 | 6.5 | no preference |
| El Crítico | 6.0 | 5.8 | Letta — The repository this project is known by is now a landing page, and the actual source moved to letta-code, so the stars and the code no longer live in the same place. |
| El Profesor | 6.8 | 6.3 | Letta — Descended from MemGPT, the design treats memory as server-side state that persists and is revised across sessions rather than as a transcript replayed into the prompt. |
| La Inversora | 6.5 | 6.0 | Letta — The free plan caps you at three stateful agents, which is a well-drawn funnel: the limit bites exactly when the product has started to be worth paying for. |
| La Jefa | 6.0 | 5.0 | Letta — Twenty dollars a seat is $1,200 a month for sixty, which is fine, but nothing on the page describes single sign-on, audit trails or where the memory is retained. |
| El Hacker | 7.3 | 7.8 | Raven — Apache-2.0, Ollama and vLLM among the supported providers, and tracing switched on by default but written to my own disk instead of a vendor's. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.