glm-5.3-flash

8/27/2026, 2:01:12 PM · 15 tests · run 12603105-7c69-4502-b160-a3d87fc08e5f

basic-generation

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

Produced 73 words (331 output tokens).

Result

{
  "wordCount": 73,
  "outputTokens": 331
}

9dfed676-2a2d-4318-af2e-6e543e91f9b3

Aug 27, 2026, 13:57:50

zai:glm-5.3-flash

Input

Output

331

Reasoning

240

Answer

Cost

$0.000169

Mode

streaming

Cost breakdown

Input $0.000003Output $0.000166

Latency

10.6s

TTFT

8.1s

Tok/s

31.3

Tok/s (adj)

37.1

usage-cost

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

in=20 out=208 rt=201 cost=$0.000107 (paid).

Result

{
  "inputTokens": 20,
  "outputTokens": 208,
  "reasoningTokens": 201,
  "costUsd": 0.000107,
  "paid": true
}

8d54fb17-1153-4c4f-aa72-293bb4da4f83

Aug 27, 2026, 13:57:59

zai:glm-5.3-flash

Input

Output

208

Reasoning

201

Answer

Cost

$0.000107

Mode

streaming

Cost breakdown

Input $0.000003Output $0.000104

Latency

8.4s

TTFT

8.3s

Tok/s

24.7

Tok/s (adj)

77.7

streaming-structure

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

mode=streaming, 66 stream events, 1 text block(s), 35 chars.

Result

{
  "mode": "streaming",
  "streamEventCount": 66,
  "textBlocks": 1,
  "textLen": 35
}

37e2be0f-7a51-4944-b856-4e3babeb675f

Aug 27, 2026, 13:58:01

zai:glm-5.3-flash

Input

Output

Reasoning

Answer

Cost

$0.000035

Mode

streaming

Cost breakdown

Input $0.000004Output $0.000031

Latency

1.8s

TTFT

1.5s

Tok/s

33.6

Tok/s (adj)

101.1

complete-stream-parity

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

stream="Paris" · complete="Paris"

Result

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

52af58a2-f0e0-4fcf-9823-8123802b046b

Aug 27, 2026, 13:58:03

zai:glm-5.3-flash

Input

Output

Reasoning

Answer

Cost

$0.000036

Mode

streaming

Cost breakdown

Input $0.000004Output $0.000032

Latency

1.9s

TTFT

1.9s

Tok/s

33.3

Tok/s (adj)

247.8

00592de1-e7c5-4713-a007-06abc4f97c98

Aug 27, 2026, 13:58:05

zai:glm-5.3-flash

Input

Output

Reasoning

Answer

Cost

$0.000030

Mode

non-streaming

Cost breakdown

Input $0.000004Output $0.000026

Latency

2.0s

Tok/s

26.0

reasoning-conformance

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

Class (from reasoning-config): always-on.
reasoningTokens=509, outputTokens=986.

Result

{
  "class": "always-on",
  "reasoningTokens": 509,
  "outputTokens": 986
}

e9699d95-0e89-4a44-9f2f-0c8227b948d7

Aug 27, 2026, 13:58:20

zai:glm-5.3-flash

Input

Output

986

Reasoning

509

Answer

477

Cost

$0.000508

Mode

streaming

Cost breakdown

Input $0.000015Output $0.000493

Latency

15.5s

TTFT

9.2s

Tok/s

63.6

Tok/s (adj)

75.4

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
}

0a1243b2-5bc7-448d-a037-6405c55f7efd

Aug 27, 2026, 13:58:25

zai:glm-5.3-flash

Input

230

Output

Reasoning

Answer

Cost

$0.000052

Mode

streaming

Cost breakdown

Input $0.000034Output $0.000017

Latency

5.3s

Tok/s

6.6

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 is 391."

Result

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

cfd551c4-996b-4132-86f7-d4c9629e6e12

Aug 27, 2026, 13:58:31

zai:glm-5.3-flash

Input

226

Output

Reasoning

Answer

Cost

$0.000050

Mode

streaming

Cost breakdown

Input $0.000034Output $0.000017

Latency

5.1s

Tok/s

6.5

354333ad-c1bb-4f83-ab59-3654d6315ce3

Aug 27, 2026, 13:58:35

zai:glm-5.3-flash

Input

269

Output

Reasoning

Answer

Cost

$0.000046

Mode

streaming

Cost breakdown

Input $0.000040Output $0.000006

Latency

4.6s

TTFT

4.6s

Tok/s

2.6

Tok/s (adj)

1564.5

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 → "## Balance Lookup The current balance for account **acme-42** is **$1"

Result

{
  "signatureType": "none",
  "tokenCaptured": false,
  "reasoned": true,
  "answered": true,
  "continuation": "## Balance Lookup\n\nThe current balance for account **acme-42** is **$1,234.00**.\n\n## Compound Growth Calculation\n\nUsing the compound interes"
}

bc14b753-8102-43a9-b709-cd5e8e26babd

Aug 27, 2026, 13:58:41

zai:glm-5.3-flash

Input

253

Output

Reasoning

Answer

Cost

$0.000073

Mode

streaming

Cost breakdown

Input $0.000038Output $0.000036

Latency

5.5s

Tok/s

12.8

b738aa3c-a6cb-4468-8c1f-8509346a3424

Aug 27, 2026, 13:59:01

zai:glm-5.3-flash

Input

336

Output

967

Reasoning

572

Answer

395

Cost

$0.000534

Mode

streaming

Cost breakdown

Input $0.000050Output $0.000483

Latency

20.3s

TTFT

13.7s

Tok/s

47.6

Tok/s (adj)

59.7

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, USA"
  },
  "validJson": true,
  "missingKeys": [],
  "bareJson": true
}

5678f723-df4a-4c72-8813-a9f3d9794295

Aug 27, 2026, 13:59:06

zai:glm-5.3-flash

Input

Output

153

Reasoning

124

Answer

Cost

$0.000087

Mode

streaming

Cost breakdown

Input $0.000010Output $0.000077

Latency

4.9s

TTFT

4.5s

Tok/s

31.2

Tok/s (adj)

74.1

summarize-memex

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

Read ~10238 input tokens, wrote 0 words (8000 output tokens) in 119311ms.

Result

{
  "wordCount": 0,
  "inputTokens": 10238,
  "outputTokens": 8000,
  "latencyMs": 119311
}

a36c4305-0c22-4bbc-98a9-f10676a1ae2a

Aug 27, 2026, 14:01:05

zai:glm-5.3-flash

Input

10,238

Output

8,000

Reasoning

7,997

Answer

Cost

$0.0055

Mode

streaming

Cost breakdown

Input $0.0015Output $0.0040

Latency

119.3s

Tok/s

67.1

truncation

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

finishReason=length, out=100 tokens.

Result

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

fb6e0ff0-1515-47e3-b7fd-3397f368ef45

Aug 27, 2026, 14:01:09

zai:glm-5.3-flash

Input

Output

100

Reasoning

100

Answer

Cost

$0.000054

Mode

streaming

Cost breakdown

Input $0.000004Output $0.000050

Latency

3.9s

Tok/s

25.8

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"
}