New Flagship Model Added: GLM-5.3-Flash

Date: 2026-08-26 / (1405-06-04)

Summary

Z.AI's new efficient flagship model, glm-5.3-flash, is now available through AvalAI for coding, agentic, reasoning, and multimodal workloads. It supports v1/chat/completions and v1/messages, with partial support for v1/responses, and launches with promotional pricing through September 9, 2026.

---

Details

Z.AI GLM-5.3-Flash

glm-5.3-flash is the first natively multimodal model in the GLM-5 series. Its mixture-of-experts architecture has 320 billion total parameters and activates 18 billion per token, combining sparse and linear attention to reduce long-context serving cost while retaining precise retrieval across a recorded 991,000-token input context.

Key Features:

• Efficient Architecture: 320B total parameters, 18B active parameters, 45 layers, and a hybrid sparse-linear attention design
• Long Context: Up to 991,000 input tokens and 128,000 output tokens on AvalAI
• Native Multimodality: Pre-trained on a 30-trillion-token multimodal corpus for joint textual, visual, and structural reasoning
• Coding and Agents: Designed for repository work, terminal tasks, tool use, visual coding, and iterative self-verification
• Developer Capabilities: Reasoning, function calling, tool choice, prompt caching, and streaming
• Open Weights: Z.AI publishes model weights for local deployment with frameworks including SGLang, vLLM, and TokenSpeed
• Best For: Cost-sensitive coding agents, long-context analysis, business documents, interface validation, and multimodal knowledge work

Benchmark Highlights Reported by Z.AI

Evaluation	GLM-5.3-Flash	GLM-5.2
Terminal-Bench 2.1	84.3	81.0
DeepSWE v1.1	63.4	46.2
Toolathlon Verified	78.4	59.9
AutomationBench v1.0.6	48.8	26.2
Agents' Last Exam	26.3	20.4
GDPval-AA v2	1773	1504

These provider-reported benchmark results are directional evidence. Evaluate the model with representative prompts, tools, media, and acceptance criteria before routing production traffic.

Endpoint Availability

Endpoint	Support	Notes
v1/chat/completions	Supported	OpenAI-compatible Chat Completions
v1/messages	Supported	Anthropic-compatible Messages API
v1/responses	Partial support	Verify required parameters, tools, and modalities before production use

---

Promotional Pricing

Prices are in USD per 1 million tokens. Promotional pricing applies through September 9, 2026.

Period	Input	Cached Input	Output
Through September 9, 2026	$0.075	$0.015	$0.25
After September 9, 2026	$0.15	$0.03	$0.50

The promotional rates are half the standard rates. Review the pricing page when forecasting workloads that may continue beyond the promotion.

---

API Request and Response Example

Request

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "glm-5.3-flash",
    "messages": [
      {
        "role": "user",
        "content": "Review this deployment plan, identify the three highest operational risks, and propose verification steps."
      }
    ]
  }'

Response

The shortened example below illustrates the standard Chat Completions structure. Token counts and cost vary with the request and generated output.

{
  "id": "chatcmpl-glm53flash-example",
  "created": 1787774400,
  "model": "glm-5.3-flash",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "The highest risks are an untested rollback path, missing dependency health gates, and incomplete observability. Run a staged canary, rehearse rollback before expansion, and require latency, error-rate, and saturation thresholds at every gate.",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 73,
    "prompt_tokens": 25,
    "total_tokens": 98,
    "completion_tokens_details": null,
    "prompt_tokens_details": {
      "audio_tokens": null,
      "cached_tokens": null,
      "text_tokens": 25,
      "image_tokens": null
    }
  },
  "estimated_cost": {
    "unit": "0.0000201250",
    "irt": 2.31,
    "exchange_rate": 114600
  }
}

---

SDK Usage Examples

Bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "glm-5.3-flash",
    "messages": [
      {
        "role": "user",
        "content": "Plan a safe, staged refactor of this service and include tests and rollback criteria."
      }
    ]
  }'

Python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.chat.completions.create(
    model="glm-5.3-flash",
    messages=[
        {
            "role": "user",
            "content": "Plan a safe, staged refactor of this service and include tests and rollback criteria.",
        }
    ],
)

print(response.choices[0].message.content)

JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.AVALAI_API_KEY,
  baseURL: "https://api.avalai.ir/v1",
});

const response = await client.chat.completions.create({
  model: "glm-5.3-flash",
  messages: [
    {
      role: "user",
      content: "Plan a safe, staged refactor of this service and include tests and rollback criteria.",
    },
  ],
});

console.log(response.choices[0].message.content);

---

Adoption Guidance

• Use the exact model ID glm-5.3-flash.
• Evaluate coding, tool-use, long-context, and multimodal behavior with representative workloads before changing production traffic.
• Confirm support for every required parameter, tool, and modality before using the partially supported v1/responses endpoint.
• Use prompt caching for repeated long prefixes and verify cached-token counts in the response usage fields.
• Account for the standard rates that take effect after September 9, 2026 when forecasting long-term cost.

---

Related Links

• Z.AI Models Documentation
• GLM-5.3-Flash Model Reference
• Chat Completions API Reference
• Reasoning Guide
• Pricing