Inkling Small
Inkling Small is a high-performance foundation model engineered by Thinking Machines, accessible with Zero Data Retention on ARMES.
Run Inkling Small
Included on FREE plan
Context Window
128K
128K tokens
Input Token Price
$0.58
per 1,000,000 tokens
Output Token Price
$1.44
per 1,000,000 tokens
Speed Rating
Very
Very Fast
Verified Frontier Benchmarks
Empirical Evaluation Results
Evaluated across standard and verified harnesses
AA Intelligence Index
reasoning25.7
AA Coding Index
coding52.9
AnalystAgent
strategy27.5%
AutomationBench
agentic4.8%
AA Briefcase
strategy912 Elo
Critpt
reasoning8.3%
EnterpriseOps Gym
agentic42.7%
Gdp Pdf
strategy11.2%
AA GDPval 2.0
strategy1191.4 Elo
GPQA Diamond
reasoning89.5%
Humanity's Last Exam (HLE)
reasoning33.3%
Lcr (1.1)
reasoning75.7%
Mmmu Pro
reasoning74.0%
Omniscience
science-8.9
SciCode
science49.7%
Tau3 Banking
finance18.8%
Terminal-Bench 2.1
coding55.1%
Terminal-Bench 4.0
coding1.0%
Openrouter GPQA Diamond
reasoning86.4%
Openrouter Tau2 Bench Airline
finance60.9%
Slop Index
reasoning19.4
Vals EMB 2.0
finance31.8%
Vals Vibe Code Bench
coding19.1%
Vals Code Migration
coding13.3%
Vals Legal Research
legal25.5%
Mcp Atlas
reasoning79.2%
Context Arena MRCR V2 (8-needle)
agentic16.6%
Vals Index (2)
reasoning31.9%
Vals Finance Agent
finance41.3%
Vals Terminal Bench 2.1 (2)
coding55.1%
Harvey LAB-AA
legal1.7%
AA Agentic Index
agentic25.0
TAU-Bench Banking
agentic18.8%
Briefcase
strategy912 Elo
Gdpval (2)
strategy1191.4 Elo
TAU-3 Banking
agentic18.8%
TAU-2 Airline
agentic60.9%
Briefcase v1.1
strategy912 Elo
GDPval-AA v2.1
strategy1191.4 Elo
AA-LCR v1.1
reasoning75.7%
AA-Omniscience Index
reasoning-8.9
Creative Writing v3
voice1491.4 Elo
Architectural Profile & Capabilities
Inkling Small delivers dependable intelligence for coding, drafting, summarization, and interactive reasoning on the FREE tier.
Architecture Type: Dense Transformer
Standard dense autoregressive transformer architecture with full-attention mechanisms.
Recommended Workloads & Primary Use Cases
General-purpose reasoning
Technical drafting
Code assistance
Fast iteration
ARMES Zero Data Retention
Calls to Inkling Small are routed through strict Zero Data Retention inference channels. Prompts and outputs are never stored, indexed, or monitored by Thinking Machines.
More Models from Thinking Machines
Compare other engines available in this laboratory.