XingChen: Xing4.0 29B A4B (free)

XingChen: Xing4.0 29B A4B (free) is an open-weight mixture-of-experts language model from XingChen, developed by China Telecom's AI team (TeleAI). It is the successor to the TeleChat series. The model has 29B total parameters with 4B active per token, which means only a subset of the network is engaged for each token generated. It is a text-only model, listed under the text category on api14.

The model accepts text as input and produces text as output. It supports a context window of 262144 tokens, and its maximum output length is 235929 tokens. That combination makes it suitable for workloads where very long documents or extended conversations need to be processed in a single request, and where the generated response itself may be lengthy. Because it is a mixture-of-experts design, the active parameter count per token is much smaller than the total, which is relevant if you are thinking about throughput and serving characteristics rather than raw parameter count alone.

Practical use cases follow directly from the long context and text-only interface. You can use it for summarising or analysing long reports, contracts, transcripts or codebases that would not fit into smaller context windows. It suits multi-turn assistants that need to retain a large amount of prior conversation, and document question answering where the source material is passed in full. The tool and structured output support also makes it a candidate for agent-style workflows, function calling, and applications that need responses in a defined JSON shape rather than free text.

You call the model through api14's OpenAI-compatible API using the model id "xingchen-agi/xing4.0-29b-a4b:free". Because the endpoint follows the OpenAI chat completions format, you can use existing OpenAI SDKs and point them at api14. The model supports a broad set of request parameters: frequency_penalty, include_reasoning, max_tokens, min_p, presence_penalty, reasoning, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k and top_p. The presence of reasoning and include_reasoning means you can request reasoning output where the model exposes it, and tools plus tool_choice and structured_outputs cover function calling and schema-constrained generation.

One thing to weigh is that this is a text-only model, so it cannot accept images or audio as input, and it does not produce them. If your application depends on multimodal input, you would need a different model. The free variant is also worth testing against your own latency and quality requirements before you commit to it in production, since the facts here describe capabilities and parameters rather than measured performance.