Skip to content
Model Pareto

K2 Think V2

MBZUAI Institute of Foundation ModelsOpen weightsartificialanalysis.ai ↗
Compare with nearest rivalsCompare with…
Input
—
Output
—
Speed
~74tok/sest.
Context
262Ktokens

Long-tail import from Artificial Analysis (2026-09-24); AA variant: K2 Think V2. Price = AA median host price.

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.

Hardware
1× B200 180GB
FP8 weights
Throughput
~2,239 tok/s
many requests batched
Price
$0.36–3.22 /M tok
busy → light use
On your own machine
Mac Studio M5 Max 128GB (int4, ~12 tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • FP8 weights (72.5B params) + 25% KV-cache headroom ≈ 91 GB
  • 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
  • ~2,239 output tok/s aggregate at batch 64 (bandwidth-bound)
  • Blended 3:1 input:output; prefill ~14,664 tok/s
  • Low = 75% utilization at the low GPU price; high = 20% at the high price

Benchmark scores

~ italic, dashed = no published score yet, estimated from related benchmarks.

Writing & chat

MMMLUEstimated
~#4 of 5 (estimated)~82.5%
  1. #1Gemini 3.1 Pro92.6%
  2. #2Gemma 4 31B88.4%
  3. #3Claude Haiku 4.583.0%
  4. ~#4K2 Think V2Estimated82.5%
  5. #4gpt-oss-120b81.3%
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Estimated
~#41 of 54 (estimated)~1363 Elo
  1. #1Claude Opus 5.51501
  2. #2Claude Fable 5.1Few votes1496
  3. #40Qwen3 Next 80B A3B Instruct1386
  4. ~#41K2 Think V2Estimated1363
  5. #41GLM-4.6VFew votes1361
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#41 of 54 (estimated)~1383 Elo
  1. #1Claude Opus 5.51509
  2. #2Claude Fable 5.11501
  3. #40Qwen3 Next 80B A3B Instruct1399
  4. ~#41K2 Think V2Estimated1383
  5. #41GLM-4.6VFew votes1378
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#19 of 26 (estimated)~1574 Elo
  1. #1GPT-6 Astra2173
  2. #2Claude Fable 5.12162
  3. #18Inkling1611
  4. ~#19K2 Think V2Estimated1574
  5. #19Gemini 3.5 Flash-Lite1559
No published score — estimated from related benchmarks

Hard reasoning

AIME (latest)Estimated
~#14 of 25 (estimated)~64.7%
  1. #1Inkling97.1%
  2. #2Muse Glimmer94.7%
  3. #13Phi-4-reasoning-plus78.0%
  4. ~#14K2 Think V2Estimated64.7%
  5. #14Mistral Large 338.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#53 of 75 (estimated)~13.5
  1. #1Claude Opus 5.557.6
  2. #2Claude Sonnet 5.556.0
  3. #52Mistral Medium 3.514.2
  4. ~#53K2 Think V2Estimated13.5
  5. #53Command A+13.1
No published score — estimated from related benchmarks

Agents

OSWorld-VerifiedEstimated
~#10 of 12 (estimated)~67.7%
  1. #1Kimi K384.8%
  2. #2Qwen3.8 27B84.3%
  3. #9Nex-N2.5-Mini71.2%
  4. ~#10K2 Think V2Estimated67.7%
  5. #10Muse Glimmer65.9%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~last of 5 (estimated)below measured range (<47.5%)
  1. #1GLM-5.3 Flash59.1%
  2. #2Kimi K358.3%
  3. #4GPT-5.547.5%
  4. ~#5K2 Think V2Estimated<47.5%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#16 of 18 (estimated)~47.1%
  1. #1Atria Dawn Preview92.5%
  2. #2GPT-6 Astra91.5%
  3. #15Mistral Medium 3.548.6%
  4. ~#16K2 Think V2Estimated47.1%
  5. #16Nemotron 3 Ultra44.4%
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#42 of 80 (estimated)~10.7%
  1. #1Claude Opus 5.569.5%
  2. #2DeepSeek V4.1 Flash68.9%
  3. #41K2 Horizon 3.7B14.7%
  4. ~#42K2 Think V2Estimated10.7%
  5. #42Qwen3.5 397B A17B6.9%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#53 of 89 (estimated)~13.4%
  1. #1Muse Spark 1.350.5%
  2. #2GLM-5.350.3%
  3. #51Qwen3.5 397B A17B13.4%
  4. ~#53K2 Think V2Estimated13.4%
  5. #53LongCat 2.013.2%
No published score — estimated from related benchmarks