Best AI Agents for Full-Stack Apps
The best LLM for building full-stack apps is Claude Fable 5, rated 1297 in Design Arena's full-stack agent arena — given a brief and a tool loop, and judged on what it actually shipped. Claude Opus 4.8 (1286) and GLM 5.2 (1276) round out the top three.
- 1
claude-fable-5ReasoningToolsJSON+159.9 intel · $10.00/M · 64 t/s1297Agent Elo - 2claude-opus-4.8ReasoningToolsJSON+155.7 intel · $5.00/M · 72 t/s1286Agent Elo
- 3
glm-5.2ReasoningToolsJSON51.1 intel · $0.760/M · 135 t/s1276Agent Elo - 4claude-sonnet-5ReasoningToolsJSON+153.4 intel · $2.00/M · 79 t/s1272Agent Elo
- 5claude-opus-4.6ReasoningToolsJSON+137.8 intel · $5.00/M · 1M ctx1263Agent Elo
- 6claude-sonnet-4.6ReasoningToolsJSON+135.9 intel · $3.00/M · 1M ctx1253Agent Elo
- 7
qwen3.7-maxReasoningToolsJSON46.0 intel · $1.48/M · 45 t/s1240Agent Elo - 8
gemini-3.5-flashReasoningToolsJSON+250.2 intel · $1.50/M · 270 t/s1235Agent Elo - 9
minimax-m3ReasoningToolsJSON+144.4 intel · $0.300/M · 113 t/s1223Agent Elo - 10
kimi-k2.7-codeReasoningToolsJSON+141.9 intel · $0.730/M · 154 t/s1219Agent Elo - 11glm-5.1ReasoningToolsJSON40.2 intel · $0.966/M · 205K ctx1209Agent Elo
- 12kimi-k2.6ReasoningToolsJSON+144.2 intel · $0.589/M · 262K ctx1199Agent Elo
- 13glm-5v-turboReasoningToolsJSON+134.5 intel · $1.20/M · 203K ctx1190Agent Elo
- 14glm-5ReasoningToolsJSON39.5 intel · $0.950/M · 205K ctx1164Agent Elo
- 15kimi-k2.5ReasoningToolsJSON+135.4 intel · $0.570/M · 262K ctx1157Agent Elo
- 16
gpt-5.5ReasoningToolsJSON+154.8 intel · $5.00/M · 1.1M ctx1128Agent Elo - 17gemini-3.1-pro-previewReasoningToolsJSON+246.5 intel · $2.00/M · 136 t/s1110Agent Elo
- 18gemini-3-flash-previewReasoningToolsJSON+2$0.500/M · 1.0M ctx1104Agent Elo
- 19gpt-5.6-terraReasoningToolsJSON+155.0 intel · $1.00/M · 83 t/s1096Agent Elo
- 20glm-4.7ReasoningToolsJSON33.7 intel · $0.400/M · 205K ctx1092Agent Elo
- 21gpt-5.2ReasoningToolsJSON+142.2 intel · $1.75/M · 81 t/s1084Agent Elo
- 22glm-4.6vReasoningToolsJSON+111.0 intel · $0.300/M · 26 t/s1073Agent Elo
- 23glm-4.6ReasoningToolsJSON23.0 intel · $0.500/M · 36 t/s1073Agent Elo
- 24gpt-5.4ReasoningToolsJSON+151.4 intel · $2.50/M · 1.1M ctx1060Agent Elo
- 25
grok-4.3ReasoningToolsJSON+137.6 intel · $1.25/M · 1M ctx1046Agent Elo
How this is ranked
Models ranked by Design Arena Elo in the full-stack agent arena — each one is given a brief and a tool loop and has to build a working application, with humans judging the results head to head. This measures something a static coding benchmark cannot: whether a model can hold a multi-file project together, recover from its own mistakes, and finish.
Frequently asked
Which LLM builds the best full-stack app?
The best LLM for building full-stack apps is Claude Fable 5, rated 1297 in Design Arena's full-stack agent arena — given a brief and a tool loop, and judged on what it actually shipped. Claude Opus 4.8 (1286) and GLM 5.2 (1276) round out the top three.
Which AI actually builds a working web app?
The best AI model for building full-stack apps is Claude Fable 5, rated 1297 in Design Arena's full-stack agent arena — given a brief and a tool loop, and judged on what it actually shipped. Claude Opus 4.8 (1286) and GLM 5.2 (1276) round out the top three.
What's a good alternative to Claude Fable 5?
Claude Opus 4.8 (1286) is the closest alternative on this metric, followed by GLM 5.2 (1276). See the full ranking above for the tradeoffs.