kimi-k3

7/19/2026, 1:04:31 AM · 15 tests · run dfc1c5f8-b5b2-4e02-99f3-0ea8d2d3b0be

basic-generation

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

Produced 116 words (150 output tokens).

Result

{
  "wordCount": 116,
  "outputTokens": 150
}

2f09570d-60f5-47a4-816e-03688abc1444

Jul 19, 2026, 01:02:28

moonshotai:kimi-k3

Input

Output

150

Reasoning

Answer

150

Cost

$0.0023

Mode

streaming

Cost breakdown

Input $0.000087Output $0.0022

Latency

6.3s

TTFT

2.2s

Tok/s

23.8

Tok/s (adj)

36.7

usage-cost

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

in=93 out=209 rt=189 cost=$0.003414 (paid).

Result

{
  "inputTokens": 93,
  "outputTokens": 209,
  "reasoningTokens": 189,
  "costUsd": 0.0034140000000000004,
  "paid": true
}

e4391a71-e2cb-4599-8f98-5d923a9c8a25

Jul 19, 2026, 01:02:38

moonshotai:kimi-k3

Input

Output

209

Reasoning

189

Answer

Cost

$0.0034

Mode

streaming

Cost breakdown

Input $0.000279Output $0.0031

Latency

10.0s

TTFT

9.8s

Tok/s

20.8

Tok/s (adj)

107.7

streaming-structure

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

mode=streaming, 64 stream events, 1 text block(s), 35 chars.

Result

{
  "mode": "streaming",
  "streamEventCount": 64,
  "textBlocks": 1,
  "textLen": 35
}

b419497c-95a5-455a-9fc1-8fbc9cef48ec

Jul 19, 2026, 01:02:43

moonshotai:kimi-k3

Input

Output

Reasoning

Answer

Cost

$0.0014

Mode

streaming

Cost breakdown

Input $0.000294Output $0.0011

Latency

4.6s

TTFT

3.6s

Tok/s

15.5

Tok/s (adj)

42.3

complete-stream-parity

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

stream="Paris" · complete="Paris"

Result

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

64337f6e-0218-4751-a9b9-be00d505e2f2

Jul 19, 2026, 01:02:45

moonshotai:kimi-k3

Input

Output

Reasoning

Answer

Cost

$0.000261

Mode

streaming

Cost breakdown

Input $0.000096Output $0.000165

Latency

2.4s

TTFT

2.4s

Tok/s

4.6

Tok/s (adj)

8088.2

81c6d17a-d83c-4d5c-9ec9-38c7603d2e57

Jul 19, 2026, 01:02:47

moonshotai:kimi-k3

Input

Output

Reasoning

Answer

Cache Read

Cost

$0.000175

Mode

non-streaming

Cost breakdown

Output $0.000165Cache read $0.000010

Latency

1.7s

Tok/s

6.5

reasoning-conformance

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

Class (from reasoning-config): toggleable.
off: out=314 · mid: out=378 (rt=148).

Result

{
  "class": "toggleable",
  "mode": "deterministic",
  "offOutputTokens": 314,
  "midOutputTokens": 378,
  "midReasoningTokens": 148
}

eb594f54-f425-47b5-8ccc-b9ca1170d915

Jul 19, 2026, 01:02:58

moonshotai:kimi-k3

Input

103

Output

314

Reasoning

Answer

314

Cost

$0.0050

Mode

streaming

Cost breakdown

Input $0.000309Output $0.0047

Latency

11.4s

TTFT

2.7s

Tok/s

27.6

Tok/s (adj)

36.0

0226a3c2-5090-4a82-a0d3-75dcfc527181

Jul 19, 2026, 01:03:12

moonshotai:kimi-k3

Input

170

Output

378

Reasoning

148

Answer

230

Cost

$0.0062

Mode

streaming

Cost breakdown

Input $0.000510Output $0.0057

Latency

14.2s

TTFT

7.9s

Tok/s

26.5

Tok/s (adj)

36.2

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
}

5838496c-3d0f-4974-9f2e-c5b512a707e2

Jul 19, 2026, 01:03:16

moonshotai:kimi-k3

Input

172

Output

Reasoning

Answer

Cost

$0.0014

Mode

streaming

Cost breakdown

Input $0.000516Output $0.000900

Latency

3.4s

Tok/s

17.6

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
}

72bdfa91-6f0c-44c8-baab-7c067fd0dcc4

Jul 19, 2026, 01:03:20

moonshotai:kimi-k3

Input

275

Output

Reasoning

Answer

Cost

$0.0023

Mode

streaming

Cost breakdown

Input $0.000825Output $0.0015

Latency

4.8s

Tok/s

20.0

8c2e83fc-a5b9-420d-afb4-905f9d4a6f1d

Jul 19, 2026, 01:03:25

moonshotai:kimi-k3

Input

367

Output

Reasoning

Answer

Cost

$0.0017

Mode

streaming

Cost breakdown

Input $0.0011Output $0.000615

Latency

4.7s

TTFT

4.4s

Tok/s

8.8

Tok/s (adj)

88.5

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 → "The current balance for account 'acme-42' is **$1,234**. ## Compound "

Result

{
  "signatureType": "none",
  "tokenCaptured": false,
  "reasoned": true,
  "answered": true,
  "continuation": "The current balance for account 'acme-42' is **$1,234**.\n\n## Compound Growth Calculation\n\n**Formula:** A = P(1 + r)ⁿ\n\nWhere:\n- P = $1,234 (p"
}

761559ea-2053-4b95-b226-29d6de55af73

Jul 19, 2026, 01:03:34

moonshotai:kimi-k3

Input

272

Output

130

Reasoning

Answer

Cost

$0.0028

Mode

streaming

Cost breakdown

Input $0.000816Output $0.0019

Latency

8.8s

TTFT

7.7s

Tok/s

14.8

Tok/s (adj)

66.1

47136706-bef5-4ff3-a7ac-77e01daf5dd2

Jul 19, 2026, 01:03:48

moonshotai:kimi-k3

Input

304

Output

386

Reasoning

Answer

386

Cost

$0.0067

Mode

streaming

Cost breakdown

Input $0.000912Output $0.0058

Latency

13.6s

TTFT

5.2s

Tok/s

28.4

Tok/s (adj)

45.8

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
}

6adbeb28-b181-44b6-9897-7646d2de2361

Jul 19, 2026, 01:03:51

moonshotai:kimi-k3

Input

Output

Reasoning

Answer

Cost

$0.000771

Mode

streaming

Cost breakdown

Input $0.000261Output $0.000510

Latency

3.4s

TTFT

2.9s

Tok/s

9.9

Tok/s (adj)

62.4

summarize-memex

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

Read ~10274 input tokens, wrote 544 words (740 output tokens) in 32257ms.

Result

{
  "wordCount": 544,
  "inputTokens": 10274,
  "outputTokens": 740,
  "latencyMs": 32257
}

b1dcfbbb-eb2e-4909-b1ec-f6271781fe28

Jul 19, 2026, 01:04:23

moonshotai:kimi-k3

Input

10,274

Output

740

Reasoning

Answer

740

Cost

$0.0419

Mode

streaming

Cost breakdown

Input $0.0308Output $0.0111

Latency

32.3s

TTFT

4.5s

Tok/s

22.9

Tok/s (adj)

26.7

truncation

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

finishReason=length, out=100 tokens.

Result

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

55a4573e-b97f-40a4-b901-8750203f59d1

Jul 19, 2026, 01:04:28

moonshotai:kimi-k3

Input

Output

100

Reasoning

Answer

100

Cost

$0.0016

Mode

streaming

Cost breakdown

Input $0.000105Output $0.0015

Latency

5.1s

TTFT

2.3s

Tok/s

19.8

Tok/s (adj)

35.7

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"
}