Support

Thinking Machines · released 2026-11-20

Inkling 2

Measured on 57 benchmarks.

Compare

20.1

#52 of 82 models

Strengths

  • Inkling 2
  • Fable 5.2 (#1)
  • Field median
bestField medianFrontierCode 2.0-AATerminal-Benchv4-AAAutomationBench-AAAA-Briefcase v1.1GDP.pdfTerminal-Bench-Science-AACritPt v2 HardSubsetAA-Omnisciencev1.1 AccuracyHLE-DiamondAA-Omnisciencev1.1Non-hallucination
  1. FrontierCode 2.0-AA15%15.1%#48 of 82
  2. Terminal-Bench v4-AA10%5.6%#53 of 82
  3. AutomationBench-AA5%41.8%#52 of 82
  4. AA-Briefcase v1.115%1263#51 of 82
  5. GDP.pdf5%16.8%#40 of 82
  6. Terminal-Bench-Science-AA15%1.4%#52 of 82
  7. CritPt v2 Hard Subset10%17.0%#57 of 82
  8. AA-Omniscience v1.1 Accuracy10%20.5%#63 of 82
  9. HLE-Diamond10%15.2%#53 of 82
  10. AA-Omniscience v1.1 Non-hallucination5%63.4%#40 of 82

Bars show where Inkling 2 sits among the 82 ranked models on each benchmark (right is best); the tick is the field median.

By category

  • Inkling 2
  • Fable 5.2 (#1)
  • Field median
p25p50p75topField medianCodingAgents & computeruseProfessional workScience & researchMathReasoning &puzzlesKnowledge &accuracyVisionSafety & security
  1. Coding10 benchmarks26%#54 of 82
  2. Agents & computer use22 benchmarks26%#56 of 82
  3. Professional work13 benchmarks29%#55 of 82
  4. Science & research6 benchmarks23%#59 of 81
  5. Math2 benchmarks30%#34 of 47
  6. Reasoning & puzzles6 benchmarks27%#51 of 80
  7. Knowledge & accuracy2 benchmarks32%#54 of 81
  8. Vision3 benchmarks43%#30 of 68
  9. Safety & security2 benchmarks37%#34 of 46

Across 66 results on every current benchmark that measured Inkling 2, not only this ranking’s. The percentage is its average standing among the 82 models in this ranking that took the same benchmarks: 100% beat them all, 0% trailed them all. A benchmark in two categories counts in both; a category needs two benchmarks.

Cost and speed

against 82 ranked models

Output speed

#6 fastest of 67

280 tok/s

Time to first token

#12 quickest of 67

1.5s

Output tokens per task

#65 fewest of 75

152k

Token price

$0.021 per 1M tokens at API prices

97.0% cache read · 2.50% fresh input · 0.00% cache write · 0.50% output

Assumed mix: no measured T3 Code usage for this model.

Not measured for Inkling 2: cost per task, time per task (no published API price yet).

Availability

API providers

3
ProviderIn / out per 1MTok/sFirst
  • PrimeIntellect$0.07 / $0.30
  • Io Net$0.08 / $0.35
  • Thinking Machines$0.15 / $0.502801.5s

OpenRouter, Artificial Analysis

This ranking’s benchmarks

10 of 10 measured

Cost per task is what a subscriber pays: the model’s plan price, a share of API price. Models no plan sells are at API price.

Agents & computer use

AutomationBench-AA

283 results
  1. #144default of 28341.8%

Leader: Gemini 4 Neon · 79.3%

Professional work

AA-Briefcase v1.1

288 results
  1. #121default of 2881263

Leader: Sonnet 5.5 · 1813

Science & research

CritPt v2 Hard Subset

160 results
  1. #118default of 16017.0%

Leader: Fable 5.2 · 76.5%

Coding

FrontierCode 2.0-AA

159 results
  1. #100default of 15915.1%

Leader: Fable 5.2 · 49.1%

Professional work

GDP.pdf

283 results
  1. #133default of 28316.8%

Leader: Gemini 4 Neon · 33.4%

Knowledge & accuracy

HLE-Diamond

197 results
  1. #117default of 19715.2%

Leader: Fable 5.2 · 80.2%

Knowledge & accuracy

AA-Omniscience v1.1 Accuracy

171 results
  1. #141default of 17120.5%

Leader: Fable 5.2 · 79.1%

Knowledge & accuracy

AA-Omniscience v1.1 Non-hallucination

171 results
  1. #59default of 17163.4%

Leader: Gemini 4 Carbon · 95.8%

Science & research

Terminal-Bench-Science-AA

170 results
  1. #120default of 1701.4%

Leader: GPT-6.2 Astra · 78.6%

Agents & computer use

Terminal-Bench v4-AA

280 results
  1. #149default of 2805.6%

Leader: Kimi K3.1 · 67.2%

All 57 benchmarks

show 47 more