← All models

K2 Think V2

open weights
Institute of Foundation Models · released 2025-12-15 · 0 offers

Context 262K tokens

Top 0 cheapest providers (Adjusted $/task)

Within the active global provider, residency and confidentiality filters.

No per-token pricing matches the active global filters.

Composite

2 of 7 inputs6 radar axes: DesignArena's two boards share one

21.9

AA CodingCoding Agent v1.4AA IntelligenceAA AgenticEpoch ECISoftware ECIDesignArenaAA Coding: percentile 23AA Intelligence: percentile 50

AA Coding 21.0Coding Agent v1.4 AA Intelligence 11.5AA Agentic Epoch ECI Software ECI DesignArena —/—

Radar: percentile among all models measured on each input; a gap means not measured.

Benchmark sheet

11 of 140 registered benchmark versions · bars show the percentile among all models measured on each benchmark.

Compare this model →

Agentic

  • AA-Briefcase234 Elo

    AA-Briefcase published 2026-09-10 · Published board — Tests multi-week professional knowledge-work projects with linked tasks and large source collections.

  • GDPval-AA v213328 Elo

    GDPval-AA v2 v2 · Published board — Tests professional knowledge-work deliverables across occupations using AA's Stirrup harness.

  • Terminal-Bench Hard (AA)376.8%

    Terminal-Bench Hard (AA) pinned revision 74221fb · Published board — Tests a pinned 44-task hard subset of terminal-based work using Terminus 2.

  • Terminal-Bench v2.1 (AA)2315.0%

    Terminal-Bench v2.1 (AA) v2.1 · Published board — Tests terminal-based work on the 89-task verified refresh using Terminus 2.

Coding

  • AA Coding Index2321.0

    AA Coding Index published 2026-09-19 · Published board — Artificial Analysis publishes this board without a version number; we keep the date each result was retained.

Instruction-following

Knowledge

Long-context

  • AA-LCR v1.15357.0%

    AA-LCR v1.1 v1.1 · Published board — Tests reasoning across multiple long documents with corrected answer keys and grading.

Reasoning

Science

Tool-use

Unusual results

1 threshold-crossing signal flagged · 13 eligible benchmark families

How flags are calculated

Heuristic screen, not statistical significance: benchmark families are correlated and source uncertainty is unknown. Peer evidence needs ≥ 20 independently measured matched configurations from ≥ 10 distinct model families. A flag needs a directed population z-score of magnitude ≥ 1.5 and a gap of ≥ 1.5 from the leave-one-benchmark-family-out mean z in the same direction, over ≥ 5 other benchmark families.

  • unusually weak AA-Briefcase published 2026-09-10
    Why
    Observed: 34 Elo
    Peer mean 916.66604 · peer sd 463.03709 · n 154 · families 108
    Directed z -1.906 · baseline z -0.406 · gap -1.5 · profile n 12
    34 Elomeasuredobserved 2026-09-10artificialanalysis.ai
    Evidence
    Axis: AA-Briefcase · published 2026-09-10 · Published board
    Exact value: 34 Elo
    Observed: 2026-09-10T21:47:16.627Z · publication date: not recorded
    Observation id: aa:5a49ef80-3af5-404b-8ac0-e1b230ae95de:briefcaseBreakdown.overall.elo
Missing coverage · 132 benchmark versions

No result does not mean a zero, or that the model was never tested. Collection failures and disputed versions retain their distinct status.

Token offers by platform · 0 offers (Adjusted $/task)

Click any underlined price to see how it is estimated and where each input comes from. How we calculate adjusted cost.

0 offers within the active global filters; “—” means the catalog is active but no public token price is available.

No token offers match the active global filters.

K2 Think V2 — benchmarks & cost | Benchmark Heaven