Skip to content
OSMEXO

Inkling Small

thinking-machines/inkling-small
  • Thinking Machines
  • Family · Inkling Small
  • Released 30 Jul 2026
  • Open weights
  • Text · Image · Audio → Text
  • Reasoning · adjustable
  • Tools · Vision
Not on OSMEXO yet
Spec & list price
Context1M
Max output256K
Input / 1M$0.45
Output / 1M$1.202.7× input
Scores · standing
Intelligence25.7#90/489
Coding52.9#60/196
Usage
Tokens / day28.7B▼ 83% 7d
Usage rank#62
Requests / day18.1K
Tool calls / req0.44
Per request · 30 days
Avg input12.5K0.8× median
Avg output1.0K1.5× median
Cache hit77%about median
Reasoning share78%1.4× median
Tool-call failures2.6%1.4× median

Consumption

2.7B5.3B8.0B3 Oct17 Oct1 Nov
Peak 7.5B 21 OctLow 110.8M 1 NovRank today #62

News & trends

News · latest from Thinking Machines

Benchmarks

bar = share of the top score · Benchmark data: Artificial Analysis
GPQA Diamond#49 of 445
89.5%
Long-context reasoning#97 of 393
75.7%
SciCode#61 of 160
49.7%
Coding index#60 of 196
52.9
Terminal-Bench 2.1#68 of 182
55.1%
Humanity's Last Exam#85 of 465
33.3%
Intelligence index#90 of 489
25.7
τ-Bench Banking#65 of 154
18.8%
Speed · median across 2 providers, latest day
Output speed#46/334126 tok/s1.9× faster than median
Fastest provider175 tok/sbest of 2 providers
First token#1751.11 sabout median
First answer token#4411 sthinking included

#10 cheapest of 28 with a similar score

Value
Intelligence Index
202224262830$0.2$0.5$1$2Inkling Small
Blended price per 1M tokens (log)

Technical details

Reasoning
on · high
Tokenizer
Other
License
Apache 2.0
Parameters
266B
Moderation
none
Downloads · likes
1.4M · 435
Reasoning effort
  • max
  • high · default
  • medium
  • low
  • minimal
  • none
Request parameters 16
  • tools
  • tool_choice
  • reasoning
  • reasoning_effort
  • include_reasoning
  • temperature
  • top_p
  • top_k
  • min_p
  • max_tokens
+6 more
  • stop
  • seed
  • frequency_penalty
  • logit_bias
  • presence_penalty
  • repetition_penalty

Interesting facts

  1. 2.5× the output room of similar models (262.1K against a 106.5K median).
  2. Weekdays average 1.4B tokens a day against 787.6M on weekends.

When to use

Best for

  • Audio and image understanding in apps
  • Graduate-level science question answering
  • Very long document analysis
  • Coding help with tool calling

Think twice for

  • Structured-output pipelines
  • Hardest agentic benchmarks
  • Image or audio generation
  • Reasoning-effort tuning

Alternatives

other vendors · similar capability

Want Inkling Small on OSMEXO?

Sign in to request it. Requests are reviewed; a request does not guarantee availability.