Skip to main content
TokenCost logoTokenCost

GLM-5.3

Last verified August 22, 2026 · Zhipu pricing ↗

$1.40/1M input · $4.40/1M output · 1.0M context · Zhipu

GLM-5.3 costs $1.40 per 1M input tokens and $4.40 per 1M output tokens from Zhipu, with a 1.0M-token context window. A typical chat reply (1,500 tokens in, 400 out) costs $0.0039, or $38.60 per 10,000 requests. On that workload it is cheaper than 75 of the 179 other buyable models tracked here.

Count Tokens

—
Tokens
—
Input Cost
—
Output Cost

Estimate Monthly Cost

Monthly Cost Estimator

Quick:
<$0.0001/mo

Pick a preset above or enter custom usage

What GLM-5.3 Costs Per Request

At $1.40 per 1M input and $4.40 per 1M output tokens, before caching or batch discounts.

WorkloadInput / output tokensPer requestPer 10,000 requests
Chat reply1,500 / 400$0.0039$38.60
RAG answer over retrieved docs8,000 / 600$0.014$138.40
Coding-agent step30,000 / 2,000$0.051$508.00

GLM-5.3 vs Similarly Priced Models

Nearest buyable models by price, with the RAG answer above (8,000 in / 600 out) priced on each.

ModelInput /1MOutput /1MContextRAG answer
GLM-5.3 Zhipu$1.40$4.401.0M$0.014
GLM-5.2 Zhipu$1.40$4.401.0M$0.014
GPT-6.1 Terra OpenAI$1.00$5.001.1M$0.011
Claude Haiku 4.5 Anthropic$1.00$5.00200K$0.011

Alternatives to GLM-5.3

Pricing Details

ZhipuAnnounced August 14, 2026 reachable only through the GLM Coding Plan over the OpenAI Chat Completions-compatible protocol; the metered per-token card landed August 18, which is also the creation date OpenRouter records. Rate card is IDENTICAL to GLM-5.2 and to GLM-5.1 on all three lines: $1.40 input, $0.26 cached input, $4.40 output, confirmed on Z.ai's own pricing table 2026-08-22. Cached input storage marked 'Limited-time Free'. THE PRICE HOLDING FLAT IS MISLEADING: Artificial Analysis (v4.1.1, read 2026-08-22) measures cost per Intelligence Index task at ~$0.68 against GLM-5.2's ~$0.44, a 55% rise at identical rates, because the model is more verbose - 170M output tokens across the suite against GLM-5.2's 140M, and 2.4x the 72M median for comparable models. Total index run cost $1,238.50 against GLM-5.2's $843.44. AA Intelligence Index 60 (#9 of 186), tying Kimi K3 (60) and behind Claude Opus 5 (63); GLM-5.2 scored 53. Output 93.2 tok/s, TTFT 1.92s on the model page (the providers page lists 23.38s, a different metric - do not mix them). NO BATCH TIER: Z.ai's pricing page lists none, the docs index carries async endpoints only for image and video generation, and docs.z.ai/api-reference/llm/batch-inference returns 404. The 50% off-peak discount is a GLM Coding Plan CREDIT mechanic only (peak = Mon-Fri 14:00-18:00 Singapore time per docs.z.ai/devpack/overview) and does not touch the metered API. WEIGHTS NOT RELEASED as of 2026-08-22: AA lists the model as proprietary with no license, and huggingface.co/zai-org's newest GLM repo is still GLM-5.2. Z.ai guided to a staged release ~2 weeks out after safety evaluation, which points at ~Aug 28 but is not a committed date; The Decoder reports the reason as restricting full access to selected security partners, phrasing we could not confirm against a Z.ai page. This matters for price: GLM-5.2 is MIT and served by 21 providers from $0.37 blended (OpenRouter $0.336/$1.056, a flat 76.0% below Z.ai list), while GLM-5.3 has exactly ONE seller at full list, so upgrading from a third-party GLM-5.2 costs 4.1667x more on both lines. Vendor-reported evals, relayed secondhand because z.ai/blog is JS-rendered and unreadable: Terminal-Bench 3.0 28.3 (GLM-5.2 4.6), ExploitBench 54.4 (24.4), AutomationBench 48.2 (26.2), CyberGym 84.5 (77.2), DeepSWE v1.1 66.9 (46.2). NOTE the Terminal-Bench figure is v3.0 and AA's index uses v2.1 - different task sets. GDPval-AA v2 is 1,769 on Z.ai's chart and 1,770 in AA's independent run (Opus 5 1,855, Kimi K3 1,668). Parameter count disputed across reports between 743B and 753B, so none stored. Estimated tokens (different tokenizer).
Input / 1M tokens
$1.4
Output / 1M tokens
$4.4
Context Window
1.0M
Max Output
131K

Price History

No price change recorded since 2026-08-24, when this site started tracking this model; it launched on 2026-08-18.

Frequently Asked Questions

Common questions about GLM-5.3 pricing and usage

How much does GLM-5.3 cost?
GLM-5.3 costs $1.4 per 1 million input tokens and $4.4 per 1 million output tokens. This pricing is set by Zhipu.
What tokenizer does GLM-5.3 use?
GLM-5.3 uses the cl100k_base tokenizer. This tokenizer has a vocabulary of ~100,000 tokens and is widely used across many models.
What is GLM-5.3's context window?
GLM-5.3 supports a context window of 1.0M tokens (1,048,576 tokens). The maximum output length is 131K tokens.
How does GLM-5.3 compare to other models?
On an 8,000-token-in, 600-token-out RAG answer, GLM-5.3 costs $0.014 per request. The nearest-priced alternatives cost: GLM-5.2 (Zhipu) $0.014, GPT-6.1 Terra (OpenAI) $0.011, Claude Haiku 4.5 (Anthropic) $0.011. The table above lists their full rates and context windows.
How do I count tokens for GLM-5.3?
Use the token counter at the top of this page or visit the homepage calculator. Paste your text and it will show the exact token count and cost for GLM-5.3 using the cl100k_base tokenizer.

Read More About GLM-5.3