Ollama:qwen3.6:35b-a3b-q4_K_M is used in those projects. Other models and APIs are used if there is no other options.
- agent
- skills
- tools
- deepagents
- MCP
- A2A
- Agent Protocol
- A2UI
- claude code
- opencode
- codex
- copilot
- langchain
- langgraph
- CrewAI
- google ADK
- genkit
- microsoft agent framework
- DSPy
- agentsea: not mature, it has dependency issue of orign, can't run https://docs.hub.agentsea.ai/advanced_tutorial
- fastgpt, javascript/typescript based, self-hosted or cloud
- coze: model via its platform
- Quantization
- Knowledge Distillation-- Large model --> Small model for a specific domain
- Parallelism
- Distributed Serving
- Query Transformation
- Query Transformation (with Parallel/Fan-out Queries)
- Reciprocal Rank Fusion (RRF)
- Query Decomposition
- HyDE
- Text chunking
- Reranking
- Leveraging metadata
- Hybrid search
- Query rewriting
- Autocut
- Context distillation
- Fine-tuning the large language model (LLM)
- Fine-tuning the embedding models
| Player | Core AI Product | Agent Framework(s) | Cloud Platform |
|---|---|---|---|
| Alibaba Group | Qwen (Tongyi Qianwen) | AgentFabric, Qwen-Agent, ModelScope Agents | Alibaba Cloud (Aliyun) |
| Baidu | ERNIE Bot | AgentBuilder, Agents (Baidu) | Baidu AI Cloud |
| ByteDance | Doubao | Coze | Volcano Engine |
| Tencent | Hunyuan | Tencent Cloud AI Agent | Tencent Cloud |
| Tsinghua University | GLM (via Zhipu AI) | MetaGPT, AgentBench, Owl | — |
| Zhipu AI | GLM-4 | Agent development tools | Zhipu API |
| Shanghai AI Lab | InternLM | XAgent | — |
| DeepWisdom | — | MetaGPT | — |
| LangGenius | — | Dify | — |
| Labring | — | FastGPT | — |