We use cookies. Cookies and browser storage keep you signed in and remember your preferences. With your permission we also measure how the site is used. You can any time. See our cookie policy.
inclusionai/ling-3.0-tinyRuns on your balance at the prices above. New accounts start with free usage.
Settings
Ling 3.0 Tiny is a text model from inclusionAI. It is a mixture-of-experts model with 1.3B active parameters out of 7.9B total. The design targets responsive agents, instruction following, and multi-turn conversations, and the model supports switchable reasoning behaviour. It sits in the text category, so it is intended for language work rather than image, audio or other media.
The model takes text as input and returns text as output. Its context window is 262144 tokens, and the maximum output is 250000 tokens, which means very long inputs and very long generations are both possible within a single request. Reasoning can be switched on or off, and when enabled the reasoning content can be surfaced separately through the include_reasoning parameter. This combination of a large context and a comparatively small active parameter count makes it a candidate for workloads where latency and throughput matter alongside long documents.
Practical uses follow from those traits. Responsive agents and tool-driven workflows fit the instruction following and tools support. Multi-turn conversation systems benefit from the long context, since prior turns and reference material can stay in the prompt. Long document summarisation, extraction and question answering are viable because the context window is large. Structured output tasks work through response_format, and classification or scoring tasks can use logprobs and top_logprobs. Reasoning-heavy prompts can be handled by enabling reasoning where a task needs step-by-step work before the final answer.
You call the model through api14's OpenAI-compatible API using the model id "inclusionai/ling-3.0-tiny". Because the interface follows the OpenAI chat completions shape, existing OpenAI SDKs and clients work by pointing the base URL at api14 and setting the model id. The request parameters supported are frequency_penalty, include_reasoning, logprobs, max_tokens, presence_penalty, reasoning, response_format, seed, stop, temperature, tools, top_logprobs and top_p. Use max_tokens to bound generation, temperature and the penalty parameters to steer style and repetition, seed for reproducible sampling, and stop to end generation on chosen sequences. Tools enables function calling, and response_format constrains the output structure.
There are cases where another model may fit better. This model handles text only, so tasks that require image or audio input or output need a different model. It is a tiny variant, so if a task demands the strongest possible reasoning or knowledge, a larger model in the same family or another provider may be the better choice. For straightforward text generation, agents and long-context work within the supported parameters, it is a reasonable default.
This community has no topics yet. Open the first one about inclusionAI: Ling 3.0 Tiny.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api14.com/api/v1",
apiKey: process.env.API14_API_KEY,
});
const result = await client.chat.completions.create({
"model": "inclusionai/ling-3.0-tiny",
"messages": [
{
"role": "user",
"content": "Explain vector databases in two sentences."
}
]
});
console.log(result.choices[0].message.content);
frequency_penaltyinclude_reasoninglogprobsmax_tokenspresence_penaltyreasoningresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_logprobstop_p