GLM-5.3-Flash Now Available on Telnyx Inference

GLM-5.3-Flash is now available on the Telnyx Inference API. It is the first native multimodal model in the GLM-5 series, delivering stronger intelligence than GLM-5.2 at a fraction of the cost. The model runs on Telnyx-owned GPU infrastructure with an introductory 50% pricing discount through September 9.

What's new

• GLM-5.3-Flash: Available with model ID zai-org/GLM-5.3-Flash. A 320B-parameter model with 18B activated parameters, using a hybrid sparse-plus-linear attention architecture. 1M token context window, MIT-licensed, hosted on Telnyx-owned GPUs.
• Native multimodal vision: The first GLM-5 model with built-in vision. Pass images via image_url content blocks in the Chat Completions API. No separate vision endpoint or model swap required.
• Hybrid architecture efficiency: Combines sparse attention and linear attention to reduce attention computation by 3.01x and KV cache size by 4.44x compared to GLM-5.3, keeping serving costs low without sacrificing long-context precision.
• Introductory pricing: Input at $0.075/1M tokens, output at $0.25/1M tokens, cached input at $0.015/1M tokens. 50% off through September 9, 2026.
• Drop-in integration: OpenAI-compatible Chat Completions API. No new endpoints or SDK updates required.

Why it matters

GLM-5.3-Flash delivers stronger intelligence than GLM-5.2 while costing roughly 92% less per input token and 94% less per output token. The hybrid architecture means that efficiency comes from the model design itself, not from running on smaller or lower-quality hardware. For teams running high-volume inference, that gap turns a model swap into real margin improvement with zero integration work. Native vision support means teams building multimodal applications can use one model for text and image tasks instead of routing between separate vision and text endpoints.

Example use cases

• AI teams running high-volume classification, summarization, or routing workloads where a capable-but-efficient model beats a flagship LLM on cost per call.
• Developers building multimodal applications that need text and image understanding in a single API call, without managing separate vision and text models.
• Voice AI teams reserving flagship models for complex reasoning steps and routing simpler agent turns to GLM-5.3-Flash, all on the same Telnyx platform.
• Engineering teams running agentic workflows where the model can observe rendered output, debug issues, and iterate on code or documents.

Getting started

1. Open the LLM Playground in Mission Control.
2. Select zai-org/GLM-5.3-Flash from the model dropdown.
3. Send a prompt to test, or integrate it into your application using the Chat Completions API.

curl https://api.telnyx.com/v2/ai/chat/completions \
  -H "Authorization: Bearer $TELNYX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zai-org/GLM-5.3-Flash",
    "messages": [
      {"role": "user", "content": "Summarize this support ticket in one sentence."}
    ]
  }'

For image input, pass an image_url content block:

curl https://api.telnyx.com/v2/ai/chat/completions \
  -H "Authorization: Bearer $TELNYX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zai-org/GLM-5.3-Flash",
    "messages": [
      {"role": "user", "content": [
        {"type": "text", "text": "What is in this image?"},
        {"type": "image_url", "image_url": {"url": "https://example.com/image.png"}}
      ]}
    ]
  }'

Learn more in the Inference API docs or on the pricing page.