GLM-5.3 benchmarks
GLM-5.3 is an open-weights model from Z.ai, released 14 Aug 2026. It ranks 14th of 51 on our overall leaderboard and 8th of 41 for agents. At A$3.02 per million tokens (blended), it costs about 1.2× cheaper than the median model we track. It writes about 63 tokens a second, and answers start after 35 s on average, thinking included. Its weights are public, so it can be hosted in Australia on your own infrastructure or an Australian cloud region.
- CONTEXT WINDOW
- 1,000,000 tokens
- MAX OUTPUT
- 128,000 tokens
- INPUT
- text
- WEIGHTS
- Open (MIT)
- RELEASED
- 14 Aug 2026
- OUTPUT SPEED
- 63 tok/s
- ANSWER STARTS AFTER
- 35 s (thinking included)
GLM-5.3 across our rankings.
Benchmark results.
Independent test scores are for the max effort setting. “Among tracked” ranks GLM-5.3 against the 51 models on this site; LMArena ranks run across its full leaderboard.
| BENCHMARK | RESULT | AMONG TRACKED | SOURCE DETAIL |
|---|---|---|---|
| INDEPENDENT TESTS · ARTIFICIAL ANALYSIS | |||
| AA Intelligence Index | 44.8 | #12 of 51 | max effort |
| AA Coding Index | 74.8 | #14 of 42 | max effort |
| GPQA Diamond | 91.7% | #24 of 44 | max effort |
| Humanity’s Last Exam | 42.3% | #26 of 51 | max effort |
| AA-LCR | 79.7% | #30 of 51 | max effort |
| Terminal-Bench 2.1 | 83.9% | #17 of 42 | max effort |
| τ²-Bench (banking) | 50.3% | #3 of 41 | max effort |
| SciCode | 59.0% | #6 of 46 | max effort |
| BLIND HUMAN VOTES · LMARENA | |||
| LMArena Text10,960 votes · ±6 | 1483 | #16 of 42 | #19 of 402 on LMArena · 13 Sep 2026 |
| LMArena WebDev5,763 votes · ±9 | 1620 | #10 of 45 | #14 of 129 on LMArena · 22 Sep 2026 |
| LMArena Agent73,287 sessions | #18 | #16 of 37 | #18 of 46 on LMArena · 15 Sep 2026 |
What GLM-5.3 costs in Australian dollars.
List prices converted at the snapshot’s RBA rate, excluding GST. Reasoning models are billed for their thinking as output tokens, so heavier settings cost more than the job estimates show.
| PER MILLION TOKENS | AUD | USD LIST |
|---|---|---|
| Input tokens | A$1.97 | US$1.40 |
| Output tokens | A$6.18 | US$4.40 |
| Blended (3 in : 1 out) | A$3.02 | US$2.15 |
| JOB | COST (AUD) | MEDIAN MODEL |
|---|---|---|
| Customer support replyPER 1,000 REPLIES | A$5.78 | ≈ A$6.58 |
| Summarise a 30-page documentPER 100 DOCUMENTS | A$4.43 | ≈ A$4.74 |
| Agentic coding taskPER 10 TASKS | A$3.44 | ≈ A$3.72 |
Running GLM-5.3 in Australia.
Its weights are public, so it can be hosted in Australia on your own infrastructure or an Australian cloud region. How the platforms compare →
| PLATFORM | AVAILABILITY | DETAIL |
|---|---|---|
| AWS Bedrock · Sydney | NOT OFFERED | |
| AWS Bedrock · Melbourne | NOT OFFERED | |
| Azure · Australia East | NOT OFFERED | |
| Google Vertex AI · Sydney | NOT OFFERED | |
| Your own infrastructure | SELF-HOST | Open weights: run it on your own servers or GPU instances in an Australian region. |
Compare it with.
GLM-5.3, answered.
How much does GLM-5.3 cost in Australian dollars?
GLM-5.3 lists at US$1.40 per million input tokens and US$4.40 per million output tokens — A$1.97 and A$6.18 at A$1 = US$0.7123 (RBA, 22 Sep 2026), excluding GST. A typical customer support reply works out at about A$5.78 per 1,000 replies, before any reasoning tokens.
Can I use GLM-5.3 in Australia with data kept onshore?
GLM-5.3’s weights are public, so it can be hosted in Australia on your own infrastructure or an Australian cloud region. Availability comes from the cloud providers’ own documentation; check your provider’s current terms before relying on it for data-residency obligations.
Is GLM-5.3 good for coding?
It ranks 11th of 42 on our coding ranking, with an Artificial Analysis Coding Index of 74.8 and an LMArena WebDev rating of 1620 (14th of 129). The current leader is Claude Fable 5.1.
How good is GLM-5.3 at agent and automation work?
It ranks 8th of 41 on our agents ranking, completing 50.3% of τ²-Bench banking customer-service tasks and 83.9% of Terminal-Bench 2.1 tasks.
How fast is GLM-5.3?
Artificial Analysis measures GLM-5.3 at about 63 tok/s of output, with the answer starting after 35 s on average once thinking time is included (max effort setting).
What is GLM-5.3’s context window?
Z.ai documents a context window of 1,000,000 tokens (1M), with up to 128,000 output tokens. On AA-LCR, which tests reasoning across ~100,000-token document sets, it scores 79.7%.
Ratings: LMArena leaderboard dataset (CC BY 4.0), rescaled for the lens scores · leaderboards to 22 Sep 2026
Evaluations, prices and speed: Artificial Analysis (artificialanalysis.ai) · fetched 23 Sep 2026
Exchange rate: Reserve Bank of Australia, table F11.1 · A$1 = US$0.7123 on 22 Sep 2026 · prices exclude GST
Snapshot 23 Sep 2026 · updated weekly · How the rankings work →