together:inkling

7/17/2026, 8:04:22 AM · 15 tests · run 8776d1b5-7c01-4a05-b073-6b61818e4a9f

basic-generation

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

Produced 99 words (128 output tokens).

Result

{
  "wordCount": 99,
  "outputTokens": 128
}

05e2f0db-c7ee-4f48-8b11-31aa58cb6f06

Jul 17, 2026, 08:03:56

togetherai:thinkingmachines/Inkling

Input

Output

128

Reasoning

Answer

128

Cost

$0.000547

Mode

streaming

Cost breakdown

Input $0.000029Output $0.000518

Latency

1.2s

TTFT

362ms

Tok/s

108.6

Tok/s (adj)

156.7

usage-cost

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

in=23 out=280 rt=269 cost=$0.001162 (paid).

Result

{
  "inputTokens": 23,
  "outputTokens": 280,
  "reasoningTokens": 269,
  "costUsd": 0.0011615999999999998,
  "paid": true
}

e207a01e-dd07-45b1-940d-cecc4ed1066d

Jul 17, 2026, 08:03:58

togetherai:thinkingmachines/Inkling

Input

Output

280

Reasoning

269

Answer

Cost

$0.0012

Mode

streaming

Cost breakdown

Input $0.000028Output $0.0011

Latency

2.5s

TTFT

2.5s

Tok/s

111.7

Tok/s (adj)

223.6

streaming-structure

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

mode=streaming, 114 stream events, 1 text block(s), 35 chars.

Result

{
  "mode": "streaming",
  "streamEventCount": 114,
  "textBlocks": 1,
  "textLen": 35
}

e320f571-09d8-461a-8100-b4b2a49c7ebd

Jul 17, 2026, 08:03:59

togetherai:thinkingmachines/Inkling

Input

Output

112

Reasoning

Answer

Cost

$0.000487

Mode

streaming

Cost breakdown

Input $0.000034Output $0.000454

Latency

1.0s

TTFT

786ms

Tok/s

110.7

Tok/s (adj)

155.1

complete-stream-parity

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

stream="Paris" · complete="Paris"

Result

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

de563066-425f-4b9e-9c14-1ff1ae900579

Jul 17, 2026, 08:03:59

togetherai:thinkingmachines/Inkling

Input

Output

Reasoning

Answer

Cost

$0.000049

Mode

streaming

Cost breakdown

Input $0.000032Output $0.000016

Latency

159ms

TTFT

141ms

Tok/s

25.2

Tok/s (adj)

221.6

33990de0-1e18-4023-9d38-378947e227cb

Jul 17, 2026, 08:03:59

togetherai:thinkingmachines/Inkling

Input

Output

Reasoning

Answer

Cost

$0.000049

Mode

non-streaming

Cost breakdown

Input $0.000032Output $0.000016

Latency

181ms

Tok/s

22.1

reasoning-conformance

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

Class (from reasoning-config): toggleable.
off: out=391 · mid: out=648 (rt=281).

Result

{
  "class": "toggleable",
  "mode": "deterministic",
  "offOutputTokens": 391,
  "midOutputTokens": 648,
  "midReasoningTokens": 281
}

0326b592-902e-4a2c-820d-096a0a986279

Jul 17, 2026, 08:04:03

togetherai:thinkingmachines/Inkling

Input

Output

391

Reasoning

Answer

391

Cost

$0.0017

Mode

streaming

Cost breakdown

Input $0.000118Output $0.0016

Latency

3.6s

TTFT

368ms

Tok/s

108.2

Tok/s (adj)

120.5

d284242b-5044-4d38-9fbe-650dc6f50f6b

Jul 17, 2026, 08:04:07

togetherai:thinkingmachines/Inkling

Input

100

Output

648

Reasoning

281

Answer

367

Cost

$0.0027

Mode

streaming

Cost breakdown

Input $0.000120Output $0.0026

Latency

4.4s

TTFT

2.0s

Tok/s

146.8

Tok/s (adj)

153.3

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
}

b0246491-3541-46cc-baa1-74ba405a9396

Jul 17, 2026, 08:04:08

togetherai:thinkingmachines/Inkling

Input

136

Output

Reasoning

Answer

Cost

$0.000240

Mode

streaming

Cost breakdown

Input $0.000163Output $0.000077

Latency

391ms

TTFT

315ms

Tok/s

48.6

Tok/s (adj)

250.7

tool-loop

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

step1 (required): multiply({"a":23,"b":17}) → step2: "The product of23 and17 is391."

Result

{
  "toolName": "multiply",
  "args": {
    "a": 23,
    "b": 17
  },
  "product": 391,
  "continuation": "The product of23 and17 is391.",
  "usesResult": true,
  "forced": true
}

10495c7f-fe2e-4a0b-8e47-95963a173daa

Jul 17, 2026, 08:04:08

togetherai:thinkingmachines/Inkling

Input

138

Output

Reasoning

Answer

Cost

$0.000425

Mode

streaming

Cost breakdown

Input $0.000166Output $0.000259

Latency

589ms

Tok/s

108.7

dff68a62-e8b9-4b12-a733-6a1c641839d2

Jul 17, 2026, 08:04:09

togetherai:thinkingmachines/Inkling

Input

213

Output

Reasoning

Answer

Cache Read

128

Cost

$0.000261

Mode

streaming

Cost breakdown

Input $0.000102Output $0.000138Cache read $0.000022

Latency

566ms

TTFT

464ms

Tok/s

60.1

Tok/s (adj)

177.0

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.

Step 1 didn't engage reasoning (adaptive discretion) — signature not exercised this run.
reasoned=false, none token —; follow-up → "The balance for **'acme-42'** is **$1,234**. Calculation (5% annual c"

Result

{
  "signatureType": "none",
  "tokenCaptured": false,
  "reasoned": false,
  "answered": true,
  "continuation": "The balance for **'acme-42'** is **$1,234**.\n\nCalculation (5% annual compound growth over7 years):\n\n\\[\n\\text{Future Value} =1234 \\times (1.0"
}

a43d7ab3-0681-4eae-9d5f-a1c712f5fdfe

Jul 17, 2026, 08:04:10

togetherai:thinkingmachines/Inkling

Input

170

Output

Reasoning

Answer

Cache Read

128

Cost

$0.000157

Mode

streaming

Cost breakdown

Input $0.000050Output $0.000085Cache read $0.000022

Latency

398ms

TTFT

263ms

Tok/s

52.8

Tok/s (adj)

155.5

622c8028-9c7f-40a6-8077-68b78e2de723

Jul 17, 2026, 08:04:12

togetherai:thinkingmachines/Inkling

Input

205

Output

192

Reasoning

Answer

192

Cost

$0.0010

Mode

streaming

Cost breakdown

Input $0.000246Output $0.000778

Latency

1.5s

TTFT

180ms

Tok/s

128.5

Tok/s (adj)

146.1

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

0bc80ae5-5736-41c5-9086-36b3c58ccbb2

Jul 17, 2026, 08:04:12

togetherai:thinkingmachines/Inkling

Input

Output

Reasoning

Answer

Cost

$0.000181

Mode

streaming

Cost breakdown

Input $0.000088Output $0.000093

Latency

286ms

TTFT

142ms

Tok/s

80.4

Tok/s (adj)

159.3

summarize-memex

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

Read ~10214 input tokens, wrote 913 words (1269 output tokens) in 9166ms.

Result

{
  "wordCount": 913,
  "inputTokens": 10214,
  "outputTokens": 1269,
  "latencyMs": 9166
}

5edb5749-191a-4757-92e3-d8f55c8a8c0d

Jul 17, 2026, 08:04:21

togetherai:thinkingmachines/Inkling

Input

10,214

Output

1,269

Reasoning

Answer

1,269

Cost

$0.0174

Mode

streaming

Cost breakdown

Input $0.0123Output $0.0051

Latency

9.2s

TTFT

505ms

Tok/s

138.4

Tok/s (adj)

146.5

truncation

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

finishReason=length, out=100 tokens.

Result

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

5b577ad6-956b-4947-933a-3d5b9989e4a8

Jul 17, 2026, 08:04:22

togetherai:thinkingmachines/Inkling

Input

Output

100

Reasoning

Answer

100

Cost

$0.000441

Mode

streaming

Cost breakdown

Input $0.000036Output $0.000405

Latency

980ms

TTFT

310ms

Tok/s

102.0

Tok/s (adj)

149.3

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"
}