Skip to content
← Leaderboard

Inkling 2 Small

Open source
Self-hostable
Free to run
Thinking Machines
Open license
text
vision
Released (8d ago)
Avg score
70.8
/ 100 · not enough data for the index
Context
1.0M
Output limit
944k
Input price
$0.30 /M
Output price
$1.00 /M
Cost per task
$1.18
Parameters
594.1B · 19.3B active
Est. VRAM (Q4)
~356.5 GB
Price history
Unchanged since Nov 13, 2026
Compare Cost estimate

Public list price, verified 23h ago · cached input $0.03 /M

Cost per task is the average spend on one task of the Artificial Analysis Intelligence Index, counting the tokens this model actually used, as of Nov 16, 2026. Wordier and longer-thinking models cost more per task than their per-token price suggests.

Added automatically from the public model catalog. Family and licence details are not curated yet; benchmark scores appear as upstream leaderboards list this model.

Where it ranks best

Scenario standing by quality score alone, among models with enough benchmark coverage for that scenario. Cost is left out so the standing does not move with price.

Benchmarks

preference

Chatbot Arena Elo
Elo

Crowdsourced pairwise human preference rankings of LLM responses. Higher Elo means more frequently preferred by users.

95% range 1391–1414 (±12) · 2,669 votes

performance

Output Speed
tok/s

Output speed in tokens per second: the 14-day median of Artificial Analysis's daily headline figure for each model, recorded nightly. Models it does not cover fall back to a dated manual figure. Higher is faster.

Time to First Token
ms

Time from request to first streamed token in milliseconds: the 14-day median of Artificial Analysis's daily headline figure for each model. Lower is snappier; reasoning models are penalised here because they think before talking.

reasoning

ARC-AGI 2
%

Second-generation ARC challenge testing fluid reasoning over abstract visual puzzles. Resists training-data memorisation by construction: each puzzle is novel and solutions require multi-step pattern induction. Frontier models are only just starting to score above chance on the harder tier.

reliability

API Success Rate
%

Uptime of the model maker's own API (standard tier), or the median across hosts when the maker does not serve it: the 14-day median of the public routing catalog's 24-hour uptime, recorded nightly.

Not yet measured: 25 benchmarks

Benchmarks that apply to this model's modalities but have no published result in the sources we track. Scores appear automatically once an upstream leaderboard lists it.

  • knowledge · MMLU-Pro, SimpleQA
  • reasoning · GPQA, HLE
  • math · MATH, AIME, FrontierMath, OTIS AIME
  • coding · HumanEval, LiveCB, Aider
  • agentic · SWE-bench, TermBench 2
  • instruction following · IFEval
  • vision · MMMU, MMMU-Pro, MathVista
  • long context · RULER
  • general · Rolling Avg
  • data analysis · Data Analysis
  • composite · Frontier
  • reliability · Stability, Format, Recovery, Safety

Reliability monitor

Loading drift signal…

Hosted endpoints

HostInput $/MOutput $/MContextSpeedTTFTUptime 24hQuant
Host 64$0.15$0.501.0M——99.6%fp4
Host 43$0.29$0.951.0M——99.9%fp4
Host 65$0.30$1.001.0M——100.0%nvfp4
Host 60$0.30$1.001.0M——96.2%fp4
Host 58$0.30$1.001.0M——97.0%unknown
Host 66$0.30$1.001.0M——100.0%unknown
Anonymised third-party hosts. Sorted by lowest output price. Per-host speed and time to first token appear where our own probe measured them; uptime is the catalog's 24-hour figure.

Compare with...