How much does GLM 5.3 cost?

On the Respan gateway, input starts at $0.07 per 1M tokens on OpenRouter, and output starts at $3.00 per 1M tokens on Inference.net, after gateway discounts. Averaged across 18 providers, it costs $1.12 per 1M input tokens and $4.76 per 1M output tokens.

Which providers serve GLM 5.3 on Respan?

18 providers serve GLM 5.3: OpenRouter, Relace, Wafer, Inference.net, DeepInfra, FriendliAI, and 10 more. Fireworks, Inceptron, Perplexity, and TogetherAI require your own provider API key (BYOK).

What is GLM 5.3's context window?

GLM 5.3 accepts up to 1.0M tokens of context on its largest provider route.

How do I use GLM 5.3 with the Respan gateway?

Point any OpenAI-compatible SDK at https://api.respan.ai/api/ with your Respan API key and set the model to "zai/glm-5.3". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

Over the last 30 days of requests through the Respan gateway, the fastest provider was OpenRouter at 584 tokens per second, and the lowest response latency was 1.7 seconds on Inference.net. Speed varies with request size and provider.

Where does GLM 5.3 rank on the Respan Index?

GLM 5.3 scores 49.4 out of 100 on the Respan Index v1.1, ranking #24 of 138 models. By category: Coding 44.6, Agents 64.1, Reasoning 38.4, Knowledge & language 42.2, and Human preference 55.8. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by llmmetric.com.

How does GLM 5.3 score on benchmarks?

GLM 5.3 has 50 published benchmark scores, including Terminal-Bench 2.1 88.2%, Toolathlon Verified 73%, and GPQA Diamond 90.9%. Each score links to where it was published; the data is collected by llmmetric.com.

GLM 5.3

GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...

Tool Use

# Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://respan.ai/docs/agent-setup.md. Use model "zai/glm-5.3".

Context window: 1.0M tokens
Input from: $0.07/M
Output from: $3.00/M
Providers: 18
Released: Aug 14, 2026

Providers

18 providers serve this model through the Respan gateway, cheapest first. Copy a model ID to call that route directly.

Provider	Context	Input	Output
OpenRouter`openrouter/z-ai/glm-5.3`	1.0M	$0.07/M	$7.00/M
Relace`relace/z-ai/glm-5.3`	1.0M	$0.12/M	$12.00/M
Wafer`wafer/GLM-5.3`	1.0M	$0.22/M	$3.39/M
Inference.net`inference_net/glm-5.3`	1.0M	$0.90/M	$3.00/M
DeepInfra`deepinfra/zai-org/GLM-5.3`	1.0M	$0.90/M	$4.00/M
FriendliAI`friendliai/zai-org/GLM-5.3`	1.0M	$1.26/M	$3.96/M
NovitaSave 10%`novita/zai-org/glm-5.3`	1.0M	$1.26/M	$3.96/M
Baseten`baseten/zai-org/GLM-5.3`	1.0M	$1.40/M	$4.40/M
EmpirioLabs`empiriolabs/glm-5-3`	1M	$1.40/M	$4.40/M
FireworksBYOK only`fireworks_ai/glm-5p3`	1.0M	$1.40/M	$4.40/M
InceptronBYOK only`inceptron/zai-org/GLM-5.3`	1.0M	$1.40/M	$4.40/M
Nebius`nebius/zai-org/GLM-5.3`	1.0M	$1.40/M	$4.40/M
Parasail`parasail/parasail-glm-53`	1.0M	$1.40/M	$4.40/M
Parasail`parasail/z-ai/glm-5.3`	1.0M	$1.40/M	$4.40/M
Parasail`parasail/zai-org/GLM-5.3`	1.0M	$1.40/M	$4.40/M
PerplexityBYOK only`perplexity/glm-5.3`	1M	$1.40/M	$4.40/M
TogetherAIBYOK only`together_ai/zai-org/GLM-5.3`	1.0M	$1.40/M	$4.40/M
Z.ai`zai/glm-5.3`	1M	$1.40/M	$4.40/M

Certified where it counts.

Built to meet the security and privacy standards that enterprise and healthcare teams require.

Pricing

Averaged across 18 providers, in USD per 1M tokens after gateway discounts.

Average input: $1.12/M
Average output: $4.76/M
Average cache read: $0.26/M
Average cache write: $0.00/M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$1.40/MBaseten, EmpirioLabs, Fireworks, Inceptron, Nebius, Parasail, Perplexity, TogetherAI, and Z.aiLowest$0.07/MOpenRouter

Throughput and latency

30-day averages per provider from requests through the Respan gateway, best first. 7 of 18 providers had traffic in that window.

Output speed

Higher is better

584tokens/sbest, on OpenRouter

OpenRouter584 tok/s
Baseten268 tok/s
Wafer160 tok/s

Time to first token

Lower is better

2.5sbest, on Baseten

Baseten2.5 s
Wafer2.6 s
OpenRouter3.4 s

Response latency

Lower is better

1.7sbest, on Inference.net

Inference.net1.7 s
Novita1.9 s
OpenRouter2.7 s

Benchmarks

50 published scores for GLM 5.3 from 12 sources. Each score links to where it was published. Percentages are out of 100. Arena ratings are Elo, marked between the lowest and highest rated model on each leaderboard.