glm-5.3

9/5/2026, 2:22:22 AM · 15 tests · run b619e6d8-82d1-434c-9031-66f958e3a44d

basic-generation

Plain streaming generation with a soft length constraint (~100 words).

Produced 91 words (1282 output tokens).

Result

{
  "wordCount": 91,
  "outputTokens": 1282
}

25d58bab-1a9d-4042-b3e3-1b7476ada7ae

Sep 5, 2026, 02:21:17

zai:glm-5.3

Input

Output

1,282

Reasoning

1,170

Answer

112

Cost

$0.0057

Mode

streaming

Cost breakdown

Input $0.000032Output $0.0056

Latency

16.8s

TTFT

16.0s

Tok/s

76.2

Tok/s (adj)

134.2

usage-cost

Asserts usage tokens populate and cost computes from the ChatReq contract.

in=20 out=245 rt=238 cost=$0.001106 (paid).

Result

{
  "inputTokens": 20,
  "outputTokens": 245,
  "reasoningTokens": 238,
  "costUsd": 0.001106,
  "paid": true
}

d86beb13-5aba-4a27-99ca-07ae47ea6807

Sep 5, 2026, 02:21:23

zai:glm-5.3

Input

Output

245

Reasoning

238

Answer

Cost

$0.0011

Mode

streaming

Cost breakdown

Input $0.000028Output $0.0011

Latency

5.1s

TTFT

5.0s

Tok/s

48.2

Tok/s (adj)

130.5

streaming-structure

Asserts a streaming call captures events that reconstruct into the final text.

mode=streaming, 80 stream events, 1 text block(s), 35 chars.

Result

{
  "mode": "streaming",
  "streamEventCount": 80,
  "textBlocks": 1,
  "textLen": 35
}

b7284f8b-4031-4a64-bb12-ef5c45f177b7

Sep 5, 2026, 02:21:24

zai:glm-5.3

Input

Output

Reasoning

Answer

Cost

$0.000365

Mode

streaming

Cost breakdown

Input $0.000035Output $0.000330

Latency

1.9s

TTFT

1.7s

Tok/s

39.5

Tok/s (adj)

135.4

complete-stream-parity

Same prompt via stream() and complete() both return valid, matching results.

stream="Paris" · complete="Paris"

Result

{
  "streamText": "Paris",
  "completeText": "Paris"
}

0c14a3d5-0200-4f7c-8129-85667a9234fc

Sep 5, 2026, 02:21:25

zai:glm-5.3

Input

Output

Reasoning

Answer

Cost

$0.000050

Mode

streaming

Cost breakdown

Input $0.000036Output $0.000013

Latency

918ms

TTFT

861ms

Tok/s

3.3

Tok/s (adj)

52.4

c4ac3e5a-2e33-47b5-8b7c-d790ef3bb20a

Sep 5, 2026, 02:21:26

zai:glm-5.3

Input

Output

Reasoning

Answer

Cost

$0.000050

Mode

non-streaming

Cost breakdown

Input $0.000036Output $0.000013

Latency

1.1s

Tok/s

2.8

reasoning-conformance

Asserts the model's reasoning knob behaves as its reasoning-config declares.

Class (from reasoning-config): always-on.
reasoningTokens=99, outputTokens=501.

Result

{
  "class": "always-on",
  "reasoningTokens": 99,
  "outputTokens": 501
}

10c2aec5-97c5-4966-819b-0438c4521706

Sep 5, 2026, 02:21:33

zai:glm-5.3

Input

Output

501

Reasoning

Answer

402

Cost

$0.0023

Mode

streaming

Cost breakdown

Input $0.000136Output $0.0022

Latency

6.7s

TTFT

1.8s

Tok/s

74.7

Tok/s (adj)

81.8

single-tool-call

Offers a calculator tool and checks the model calls it for an arithmetic question.

Called "multiply" with args {"a":23,"b":17}.

Result

{
  "toolCalled": true,
  "toolName": "multiply",
  "args": {
    "a": 23,
    "b": 17
  },
  "correct": true
}

b3d667f7-d427-495b-be55-4b84a76f3da4

Sep 5, 2026, 02:21:38

zai:glm-5.3

Input

230

Output

Reasoning

Answer

Cost

$0.000397

Mode

streaming

Cost breakdown

Input $0.000322Output $0.000075

Latency

5.3s

Tok/s

3.2

tool-loop

Forces a tool call, feeds the result back, and checks the continuation uses it.

step1 (required): multiply({"a":23,"b":17}) → step2: "23 multiplied by 17 equals 391."

Result

{
  "toolName": "multiply",
  "args": {
    "a": 23,
    "b": 17
  },
  "product": 391,
  "continuation": "23 multiplied by 17 equals 391.",
  "usesResult": true,
  "forced": true
}

e5705cb9-2bec-4f3d-9303-4e6afac7cb60

Sep 5, 2026, 02:21:43

zai:glm-5.3

Input

226

Output

Reasoning

Answer

Cost

$0.000488

Mode

streaming

Cost breakdown

Input $0.000316Output $0.000172

Latency

4.3s

Tok/s

9.0

6ef02a3a-3ce9-4588-a787-e573e052ff35

Sep 5, 2026, 02:21:47

zai:glm-5.3

Input

275

Output

Reasoning

Answer

Cache Read

192

Cost

$0.000219

Mode

streaming

Cost breakdown

Input $0.000116Output $0.000053Cache read $0.000050

Latency

4.0s

TTFT

4.0s

Tok/s

3.0

Tok/s (adj)

164.1

web-search

Enables server-side web search and checks invocation counts, the per-request cost surcharge, and provider-executed output blocks.

Skipped: model does not support the catalog's web_search tool version.

Result

{
  "skipped": "no server-side web search support"
}

reasoning-tool-continuity

Reasoning model: continuity token captured on a tool call and replayed so the follow-up succeeds.

reasoned=true, none token —; follow-up → "The current balance for **acme-42** is **$1,234.00**. **Calculation w"

Result

{
  "signatureType": "none",
  "tokenCaptured": false,
  "reasoned": true,
  "answered": true,
  "continuation": "The current balance for **acme-42** is **$1,234.00**.\n\n**Calculation with 5% annual compounding over 7 years:**\n\nBalance after 7 years = 123"
}

7697d74b-915e-4c19-852d-41bebc9e4f78

Sep 5, 2026, 02:21:55

zai:glm-5.3

Input

253

Output

Reasoning

Answer

Cache Read

192

Cost

$0.000241

Mode

streaming

Cost breakdown

Input $0.000085Output $0.000106Cache read $0.000050

Latency

3.9s

Tok/s

6.1

e790c3e2-998f-430b-8cc8-a65c08a31787

Sep 5, 2026, 02:22:02

zai:glm-5.3

Input

289

Output

225

Reasoning

Answer

225

Cost

$0.0014

Mode

streaming

Cost breakdown

Input $0.000405Output $0.000990

Latency

6.6s

TTFT

3.7s

Tok/s

34.1

Tok/s (adj)

77.2

prompt-cache

Repeat a large prefix; assert cache-read tokens are reported and cost drops.

Skipped: provider has no (known) prompt-cache reporting.

Result

{
  "skipped": "cacheStyle=none"
}

json-output

Requests a strict JSON object and checks it parses with the required keys.

Output parsed as JSON.
All required keys present.
Output is bare JSON (no leading prose/fence).

Result

{
  "parsed": {
    "name": "Anthropic",
    "founded": 2021,
    "headquarters": "San Francisco, California, United States"
  },
  "validJson": true,
  "missingKeys": [],
  "bareJson": true
}

ed338a8a-3047-4ff2-9f83-784cf99af74a

Sep 5, 2026, 02:22:04

zai:glm-5.3

Input

Output

Reasoning

Answer

Cost

$0.000230

Mode

streaming

Cost breakdown

Input $0.000098Output $0.000132

Latency

2.1s

TTFT

1.7s

Tok/s

14.6

Tok/s (adj)

75.8

summarize-memex

Summarizes an ~11k-token essay to ~500 words (long input, reasoning minimized).

Read ~10238 input tokens, wrote 563 words (754 output tokens) in 14148ms.

Result

{
  "wordCount": 563,
  "inputTokens": 10238,
  "outputTokens": 754,
  "latencyMs": 14148
}

1c8d0997-a537-41c9-8546-223c531839c7

Sep 5, 2026, 02:22:18

zai:glm-5.3

Input

10,238

Output

754

Reasoning

Answer

751

Cost

$0.0177

Mode

streaming

Cost breakdown

Input $0.0143Output $0.0033

Latency

14.1s

TTFT

1.5s

Tok/s

53.3

Tok/s (adj)

59.3

truncation

Tiny maxTokens on a long prompt yields finishReason:"length".

finishReason=length, out=100 tokens.

Result

{
  "finishReason": "length",
  "outputTokens": 100
}

9d6b40e4-7b3a-4ed2-afa6-97816a94a35e

Sep 5, 2026, 02:22:20

zai:glm-5.3

Input

Output

100

Reasoning

Answer

100

Cost

$0.000481

Mode

streaming

Cost breakdown

Input $0.000041Output $0.000440

Latency

2.7s

TTFT

963ms

Tok/s

37.4

Tok/s (adj)

58.5

streaming-only-rejection

Streaming-only models reject complete(); all others accept it.

streamingOnly=false; complete() succeeded.

Result

{
  "streamingOnly": false,
  "completeThrew": false
}

abort

A pre-aborted signal rejects the call (abort is threaded + honored).

pre-aborted signal → rejected (AbortError).

Result

{
  "rejected": true,
  "errorName": "AbortError"
}