Glm 5.3 Flashx Legal
Neural Network
GLM-5.3-FlashX-Legal by Z.ai is a legal branch of the multimodal model for analyzing legal documents.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 048 576 tokens
- Max answer length
- 131 072 tokens
- Reasoning effort
- low, high, max
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
import OpenAI from 'openai';
const openai = new OpenAI({
apiKey: '<your bothub access token>',
baseURL: 'https://openai.bothub.chat/v1'
});
// Sync - Text generation
async function main() {
const chatCompletion = await openai.chat.completions.create({
messages: [{ role: 'user', content: 'Say this is a test' }],
model: 'glm-5.3-flashx-legal',
});
}
// Async - Text generation
async function main() {
const stream = await openai.chat.completions.create({
messages: [{ role: 'user', content: 'Say this is a test' }],
model: 'glm-5.3-flashx-legal',
stream: true
});
for await (const chunk of stream) {
const part: string | null = chunk.choices[0].delta?.content ?? null;
}
}
main();
How it works Glm 5.3 Flashx Legal?
GLM-5.3-FlashX-Legal by Z.ai is a legal branch of the GLM-5.3-FlashX multimodal model: it accepts text, images, and video, outputting text. It differs from the base version through additional training on legal data, including mid-training and reinforcement post-training, making it more confident in legal tasks. Via BotHub, the model is available without VPN or foreign cards, payment is in rubles on a pay-as-you-go basis, and all models open in one window, so switching doesn't require rewriting integrations: a unified OpenAI-compatible API is convenient for multimodal systems, while AES-GCM encryption and no data storage are suitable for sensitive documents. Lawyers can use it to check contracts and prepare opinions, compliance teams to reconcile internal policies with requirements, analysts to extract terms from scans and photos of documents, support teams for standard responses based on regulations, and legal service developers as an engine under the hood.Frequently asked questions about Glm 5.3 Flashx Legal
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
The model works with text and documents, accepts images as input, can reason before answering, and call functions. A context of 1,048,576 tokens allows handling large volumes of material, making it suitable for analyzing documentation, long files, analytics, and code.
In a single response, the model outputs up to 131,072 tokens — enough for a comprehensive analysis or a long code snippet. If the task doesn't fit in one response, you can break it down: the 1,048,576-token context allows keeping all material in the dialogue.
Yes, reasoning before answering is supported — this is noted in the model's specifications. This mode is useful for step-by-step tasks: analyzing terms, checking logic, and working with code, where it is important to think through the solution path first and then provide the final result.
Yes, both capabilities are confirmed: the model supports function calling and structured JSON output. This means it can be connected to external services and databases, and you can receive responses in a format that your code can parse immediately.
The model accepts text, images, and documents as input — this is stated in the data. We have no notes regarding audio and video, so we will not promise their support. If you specifically need these formats, test them in practice before building a workflow around them.
There is no note about the level of Russian language proficiency in the model data, so we make no specific promises. The model accepts text queries and documents via BotHub, and the easiest way to check how accurately it works with Russian is to test it with your own examples and texts.