LegionvsYoutu-Agent
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
Legion
Software Mansion · Agent framework
OSS
Panel
6.52 spec wins
- Reliability
- 6.0
- Usefulness
- 6.2
- Cost
- 8.0
- Longevity
- 5.8
“It writes Lua to get work done in Elixir, which is two functional languages agreeing that a third one should take the blame.”
Youtu-Agent
Tencent Cloud · Agent framework
OSS
Panel
6.64 spec wins
- Reliability
- 5.7
- Usefulness
- 6.5
- Cost
- 8.3
- Longevity
- 6.0
“It ships a reinforcement-learning pipeline for training your agent, so the agent now has homework and you have a GPU bill.”
Spec by spec
| Spec | Legion | Youtu-Agent |
|---|---|---|
| Architecture | ||
| Category | Agent framework | Agent framework |
| Runs | local | local |
| Platforms | macos, linux | macos, linux |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | No | No |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | No | No |
| Multi-file edits | Yes | No |
| Git operations | No | No |
| Browser control | No | Yes |
| Sandboxed execution | YesLegion runs generated Lua or Elixir in its own sandbox; the README does not describe a container runtime. | No |
| Multi-agent orchestration | No | Yes |
| Headless / CI mode | No | No |
| Models | ||
| Backbone | multiple providers | DeepSeek, gpt-oss, any OpenAI-compatible endpoint |
| Bring your own model | Yes | Yes |
| Local models | No | YesThe framework is explicitly optimised for self-hosted open-weight models, and the companion Youtu-Tip macOS app runs offline models via Ollama. |
| Cost | ||
| Pricing model | byok | byok |
| Starts at | $0/mo | $0/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | MIT | Tencent uTu-agent License |
| GitHub stars | 220 | 4,622 |
Which one would each critic pick
| Critic | Legion | Youtu-Agent | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.8 | 7.0 | Youtu-Agent — Pick Youtu-Agent if you want serious agent behaviour out of open weights; pick Agno if you would rather build against frontier models and not think about it. |
| El Crítico | 6.0 | 6.3 | Youtu-Agent — It generates tool code from a description and then executes it, and an experience-learning module changes behaviour over time, so the system you tested is not the one you run. |
| El Profesor | 7.8 | 6.8 | Legion — One evaluation can filter, branch and loop, so a sequence that costs a tool-calling agent a model round trip per step collapses into a single call. The claim is structural and checkable. |
| La Inversora | 6.3 | 7.3 | Youtu-Agent — A cloud vendor funding proof that open weights are sufficient is buying leverage against the frontier labs whose inference it would otherwise resell. |
| La Jefa | 5.5 | 5.5 | no preference |
| El Hacker | 6.8 | 7.0 | Youtu-Agent — The weights can be mine and run offline, and the licence is the vendor's own rather than a recognised one, which is the thing that stops me shipping it. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.