Skip to content
Model Pareto

INTELLECT-3

Prime IntellectOpen weightsartificialanalysis.ai ↗
Input
—
Output
—
Speed
~94tok/sest.
Context
131Ktokens

Long-tail import from Artificial Analysis (2026-09-24); AA variant: INTELLECT-3 (based on GLM-4.5-Air). 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
2× H200 141GB
FP8 weights
Throughput
~7,235 tok/s
many requests batched
Price
$0.12–0.97 /M tok
busy → light use
On your own machine
Mac Studio M5 Max 128GB (int4, ~72 tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • FP8 weights (106B params, 12B active per token) + 25% KV-cache headroom ≈ 133 GB
  • 2× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
  • ~7,235 output tok/s aggregate at batch 260 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~77,923 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

Hard reasoning

Agents

~#12 of 15 (estimated)~55.8%
  1. #1Kimi K384.8%
  2. #2Qwen3.8 27B84.3%
  3. #11Muse Glimmer65.9%
  4. ~#12INTELLECT-3Estimated55.8%
  5. #12Nemotron 3.5 Super VL NVFP452.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
  1. #1GLM-5.3 Flash59.1%
  2. #2Kimi K358.3%
  3. ~#3INTELLECT-3Estimated53.8%
  4. #3GPT-5.6 Terra50.2%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#21 of 22 (estimated)~42.8%
  1. #1Atria Dawn Preview92.5%
  2. #2GPT-6 Astra91.5%
  3. #20Nemotron 3 Ultra44.4%
  4. ~#21INTELLECT-3Estimated42.8%
  5. #21Mistral Small 421.3%
No published score — estimated from related benchmarks
~#58 of 92 (estimated)~720 Elo
  1. #1Claude Opus 5.51866
  2. #2Muse Wonder Full1857
  3. #57Ring-2.6-1T734
  4. ~#58INTELLECT-3Estimated720
  5. #58KAT Coder Pro V2720
No published score — estimated from related benchmarks
~#50 of 85 (estimated)~6.2%
  1. #1Claude Fable 5.276.4%
  2. #2Muse Wonder Full71.3%
  3. #49Mistral Medium 3.56.3%
  4. ~#50INTELLECT-3Estimated6.2%
  5. #50GPT-5.4 nano6.0%
No published score — estimated from related benchmarks
~#63 of 91 (estimated)~10.5%
  1. #1Muse Spark 1.350.5%
  2. #2GLM-5.350.3%
  3. #62Devstral 210.5%
  4. ~#63INTELLECT-3Estimated10.5%
  5. #63Nemotron 3 Super 120B A12B10.3%
No published score — estimated from related benchmarks