Thinking Machines Labopen-weight
Released: 2026-11-05Artificial AnalysisInkling 2 Small
Inkling 2 Small from Thinking Machines, tracked on the Artificial Analysis Intelligence Index.
Intelligence Index
30.6/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
88tok/sec
Output streaming throughputTime to First Answer
69s
Time to first answer token, including any thinkingCost to Run the Index
1.18USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of Inkling 2 Small. Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- 30.6
- Cost to run the Index
- $1.18
- Time per task
- 28.5 min
- Omniscience index
- 16.1
- Reasoning model
- Yes
- Cached-input price
- $0.03 / 1M
Reasoning & knowledge
- AA-Omniscience Accuracy33.8%
- CritPt v2 (Hard)35.0%
- HLE-Diamond28.7%
- Humanity's Last Exam44.1%
Coding & terminal agents
- FrontierCode-AA 2.025.4%
- SciCode52.3%
- Terminal-Bench 2.181.3%
- Terminal-Bench 4.013.1%
- Terminal-Bench-Science 0.15.7%
Agentic work
- AA-AnalystAgent51.2%
- AA-Briefcase v1.1 (Elo)1527 Elo
- AutomationBench-AA45.6%
- EnterpriseOps-Gym-AA46.8%
- GDPval-AA (normalized Elo)1630 Elo
- τ³-Banking40.4%
Long context & documents
Health & safety
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Reasoning · CritPt v2 (Hard)35.0%
Coding & agents · Terminal-Bench 4.013.1%
Long context · AA-LCR83.3%
Knowledge · Humanity's Last Exam44.1%
Sourced Benchmark Evaluations (19)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| HLE-Diamond | Active | 28.7% | 93 | 2026-11-15 | independent | Source ↗ |
| VIALS-AA | Active | 42.9% | 93 | 2026-11-15 | independent | Source ↗ |
| EnterpriseOps-Gym-AA | Active | 46.8% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-AnalystAgent | Active | 51.2% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 45.6% | 93 | 2026-11-15 | independent | Source ↗ |
| MLCR-AA (Medical Long Context Reasoning) | Active | 20.6% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 25.2% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode-AA 2.0 | Active | 25.4% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Non-Hallucination) | Active | 73.3% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Accuracy) | Active | 33.8% | 93 | 2026-11-15 | independent | Source ↗ |
| Humanity's Last Exam | Active | 44.1% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt v2 (Hard) | Active | 35% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-LCR (Long Context Reasoning) | Active | 83.3% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench-Science 0.1 | Active | 5.7% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 4.0 | Active | 13.1% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 2.1 | Active | 81.3% | 93 | 2026-11-15 | independent | Source ↗ |
| τ³-Banking | Active | 40.4% | 93 | 2026-11-15 | independent | Source ↗ |
| SciCode | Active | 52.3% | 93 | 2026-11-15 | independent | Source ↗ |
| MMMU-Pro | Active | 81.6% | 93 | 2026-11-15 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run thinking-machines-inkling-2-small
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "thinking-machines-inkling-2-small",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for Inkling 2 Small represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.