GLM-5.3 uses the same base model as GLM-5.2 — every gain comes from post-training. Compared with GLM-5.2, it is much better at complex coding and long-horizon tasks
Fine-tuningDocs | GLM-5.3 can be customized with your data to improve responses. Fireworks uses LoRA to efficiently train and deploy your personalized model |
ServerlessDocs | GLM-5.3 is available via Fireworks' serverless API, where you pay per token. There are several ways to call the Fireworks API, including Fireworks' Python client, the REST API, or OpenAI's Python client. |
On-demand DeploymentDocs | On-demand deployments allow you to use GLM-5.3 on dedicated GPUs with Fireworks' high-performance serving stack with high reliability and no rate limits. |
Run queries immediately, pay only for usage
GLM-5.3 is Z.ai’s flagship model, released August 18, 2026. It shares GLM-5.2's base model, with all gains coming from post-training, chiefly in complex coding and long-horizon tasks.
Agentic coding, terminal automation, and vulnerability research (input is text-only). Z.ai reports a 50% gain over GLM-5.2 on Z.ai Code Bench and top open-source results on Terminal Bench 3.0.
The config.json max_position_embeddings is 1,048,576 tokens, matching Z.ai's stated 1M-token context window.
1040k tokens, available on both serverless and on-demand deployments.
Yes. The checkpoint ships natively in FP8, e4m3 format with dynamic activation scaling.
Z.ai sets temperature 1.0 and top_p 0.95 as the model's defaults. When temperature is unset, Fireworks applies those generation_config.json values. See documentation for text models.
Fireworks defaults max_tokens to 2,048 and allows generation up to the full context window. If prompt plus max_tokens exceeds the context length, max_tokens is lowered to fit rather than erroring. Z.ai caps output at 128K on its own API.
Yes to both. Streaming runs through the stream parameter, and the model page lists function calling as supported.
GLM-5.3 has 743B total parameters in a Mixture-of-Experts architecture.
Yes. Fireworks supports LoRA fine-tuning for GLM-5.3.
Fireworks bills input, cached input, and output tokens separately, priced at $1.40, $0.26, and $4.40 per 1M tokens.
Serverless ceilings for this model's size tier default to 43.2M total prompt TPM, 10.8M uncached prompt TPM, and 432k generated TPM, adaptive per account and model. On-demand deployments carry no rate limits.
Fireworks announces serverless model deprecations in advance, following its serverless model lifecycle policy. For long-term version stability it recommends on-demand deployments.
The GLM-5.3 License, an MIT-style permissive grant allowing commercial use, deployment, and fine-tuning. Licensees running a Model-as-a-Service business whose revenue exceeds $10 billion over any 12 months must pass Z.ai's security review.
No. Fireworks operates zero data retention by default and does not log prompt or generation data for open models without opt-in. The Responses API is the exception, retaining data 30 days when store is true.
