AI-generated Answer for trinhthiquynhanh
Here are two options documented by ElevenLabs that are suited for complex prompts or multi‑step conversation goals:
- GLM‑4.5‑Air (ElevenLabs‑hosted): Described as delivering top‑tier reasoning accuracy and strong tool‑calling performance for Agents, at lower cost/latency due to co‑location with ElevenLabs’ speech stack. (elevenlabs.io)
- meta‑llama/Meta‑Llama‑3.1‑70B‑Instruct‑Turbo (via Together AI integration): ElevenLabs’ Together AI guide recommends using at least this model for Agents when you need tool use and structured outputs, which aligns with complex conversational goals. (elevenlabs.io)
Reliability tip (to avoid “Fail to generate a response during the call”): Enable LLM Cascading so the platform automatically falls back to backup models on API errors, timeouts, or empty responses. You can also query the List LLMs endpoint to see what models (and capabilities) are available for your workspace/region. (elevenlabs.io)
Sources:
ElevenLabs-hosted LLMs now available in ElevenAgents
Together AI | ElevenLabs Documentation
LLM Cascading | ElevenLabs Documentation
cold nymph
twin sand