Thinking Machines: Inkling
by Thinkingmachinestext+image+audio->text4 endpoints
Neura Intelligence Index
71.3/ 100
Rank
#78—
Confidence
mediumBenchmarks
4(10% coverage)
Overview
Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...
Capabilities
Text generation
Image understanding
Audio processing
Tool use / Function calling
Extended reasoning
Prompt caching
Long context
Top-P sampling
Stop sequences
Deterministic seed
Modalities
Input
TextImageAudio
Output
Text
Technical Specifications
Context Window
1.0M tokens
Max Output
471.9K tokens
Tokenizer
Other
Uptime (24h)
9992.7%
Parameters
975000000000B
License
apache_2_0Open Weights
Country
🇺🇸US
Release Date
July 21, 20262mo
Supported Parameters
frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyseedstoptemperaturetool_choicetoolstop_ktop_p
Benchmark Performance
coding
| Benchmark | Score | Source |
|---|---|---|
| SWE-bench Verified | 77.6 | verified |
multimodal
| Benchmark | Score | Source |
|---|---|---|
| CharXiv Reasoning | 78.1 | verified |
| MMMU-Pro | 73.5 | verified |
agent
| Benchmark | Score | Source |
|---|---|---|
| MCP Atlas | 76 | verified |
Performance Over Time
Pricing
Per 1M tokens
Input$1.00
Output$4.05
Cache read$0.17
Added
July 17, 2026
Last synced 9/19/2026