Enter opens the highlighted result. Esc closes.
T

Thinking Machines Inkling 2 Small at launch

Thinking Machines released Inkling 2 Small on November 9, 2026. As of November 14, 2026, it ranks #63 of 118 models on the LLMBenchmark Index, between OpenAI GPT-5.6 Luna (#62) and Meta Muse Spark 1.2 (#64). Thinking Machines's own announcement reports 85 benchmark results, including Design Arena: game dev (1424) and ARC-AGI-2 (81.7%). It costs $0.30 per million input tokens and $1.00 per million output tokens on OpenRouter.

Benchmarks and pricing API providers

What Thinking Machines reported at launch

From “Inkling 2 Small”, November 9, 2026. The maker's own numbers, not re-run by us.

Show all 85

Independent results so far

51 scores from 51 boards that test models themselves. Ranks compare Inkling 2 Small with every model on the same board.

Coding (14)

BenchmarkScoreRankSettingSource
SWE-Bench Pro V2 97.7% #5 of 17 — Scale SEAL leaderboards
SWE-bench Multimodal (Vals AI) 43.1% #12 of 20 — Vals AI boards
CyberBench (Vals AI) 72.7% #14 of 41 — Vals AI boards
LMArena WebDev: HTML 1661 #14 of 104 — LMArena (Hugging Face dataset)
ProgramBench (Vals AI) 0.5% #18 of 61 — Vals AI boards
LMArena WebDev 1647 #19 of 105 — LMArena WebDev
LMArena WebDev: React 1660 #19 of 97 — LMArena (Hugging Face dataset)
FrontierCode 2.0 25.4% #20 of 55 — Artificial Analysis
Show all 14
Vibe Code Bench (Vals AI) 81.5% #29 of 105 OpenHands Vals AI boards
Terminal-Bench 4 (Vals AI) 12.6% #40 of 57 Mini-SWE-agent Vals AI boards
Terminal-Bench 4.0 (Artificial Analysis run) 13.1% #41 of 111 — Artificial Analysis
SciCode 52.3% #43 of 109 — Artificial Analysis
LMArena Text: coding 1455 #66 of 151 — LMArena (Hugging Face dataset)
LMArena Text: software and IT 1443 #77 of 151 — LMArena (Hugging Face dataset)

General (14)

BenchmarkScoreRankSettingSource
SAGE (Vals AI) 88.6% #8 of 28 — Vals AI boards
Long context reasoning (AA-LCR) 83.3% #14 of 176 — Artificial Analysis
GDPval-AA 1630 #19 of 117 — Artificial Analysis
Vals Index 56.1% #21 of 34 — Vals Index
LMArena Text: instruction following 1408 #77 of 151 — LMArena (Hugging Face dataset)
LMArena Text: English 1418 #86 of 151 — LMArena (Hugging Face dataset)
LMArena Text: longer queries 1403 #88 of 151 — LMArena (Hugging Face dataset)
LMArena Text: non-English 1390 #89 of 151 — LMArena (Hugging Face dataset)
Show all 14
LMArena Text: multi-turn 1408 #90 of 151 — LMArena (Hugging Face dataset)
LMArena Text: writing and language 1381 #90 of 151 — LMArena (Hugging Face dataset)
LMArena Text 1403 #91 of 151 — LMArena (Hugging Face dataset)
LMArena Text: entertainment and media 1359 #94 of 151 — LMArena (Hugging Face dataset)
LMArena Text: creative writing 1348 #95 of 151 — LMArena (Hugging Face dataset)
LMArena Text: Russian 1377 #99 of 151 — LMArena (Hugging Face dataset)

Agentic work (9)

BenchmarkScoreRankSettingSource
Harvey's Legal Agent Benchmark (Vals AI) 21.7% #5 of 34 — Vals AI boards
MCP Atlas 85.4% #5 of 35 — Scale SEAL leaderboards
Tax Agent Bench (Vals AI) 38.9% #11 of 77 — Vals AI boards
BioMysteryBench (Vals AI) 70.0% #14 of 29 — Vals AI boards
Vals Science Index 32.7% #16 of 23 — Vals AI boards
Legal Research Bench (Vals AI) 46.6% #17 of 83 — Vals AI boards
SREBench (Vals AI) 1.9% #22 of 33 — Vals AI boards
Terminal-Bench-Science (Vals AI) 2.9% #32 of 41 Mini-SWE-agent Vals AI boards
Show all 9
APEX-Agents 46.8% #39 of 59 — Epoch AI benchmarking hub

Reasoning (9)

BenchmarkScoreRankSettingSource
ARC-AGI 96.5% #13 of 75 xHigh Epoch AI benchmarking hub
MysteryMechanism (Vals AI) 25.2% #16 of 28 — Vals AI boards
ARC-AGI-2 81.7% #19 of 76 xHigh Epoch AI benchmarking hub
HLE-Diamond (Artificial Analysis run) 28.7% #21 of 74 — Artificial Analysis
CritPt v2, hard subset (Artificial Analysis run) 35.0% #27 of 56 — Artificial Analysis
Humanity's Last Exam (Artificial Analysis run) 44.1% #28 of 187 — Artificial Analysis
LMArena Text: expert 1478 #44 of 150 — LMArena (Hugging Face dataset)
LMArena Text: hard prompts 1421 #85 of 151 — LMArena (Hugging Face dataset)
Show all 9
LMArena Text: hard prompts in English 1424 #87 of 151 — LMArena (Hugging Face dataset)

Knowledge (3)

BenchmarkScoreRankSettingSource
AA-Omniscience 16.1 #21 of 58 — Artificial Analysis
LMArena Text: business and finance 1415 #75 of 151 — LMArena (Hugging Face dataset)
LMArena Text: science 1409 #97 of 151 — LMArena (Hugging Face dataset)

Intelligence (1)

BenchmarkScoreRankSettingSource
AA Intelligence Index 30.6 #23 of 55 — Artificial Analysis (via OpenRouter)

Math (1)

BenchmarkScoreRankSettingSource
MathArena 42.4% #19 of 72 — MathArena