Skip to content
Model Pareto

DiffusionGemma 26B A4B

Google DeepMindOpen weightsartificialanalysis.ai ↗
Compare with nearest rivalsCompare with…
Input
—
Output
—
Speed
~89tok/sest.
Context
256Ktokens

Long-tail import from Artificial Analysis (2026-09-24); AA variant: DiffusionGemma 26B A4B. 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
BF16 weights
Throughput
~10,963 tok/s
many requests batched
Price
$0.062–0.56 /M tok
busy → light use
On your own machine
RTX 5090 32GB (int4, 300+ tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • BF16 weights (25.2B params, 3.8B active per token) + 25% KV-cache headroom ≈ 63 GB
  • 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
  • ~10,963 output tok/s aggregate at batch 256 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~139,885 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
~#3 of 5 (estimated)~83.5%
  1. #1Gemini 3.1 Pro92.6%
  2. #2Gemma 4 31B88.4%
  3. ~#3DiffusionGemma 26B A4BEstimated83.5%
  4. #3Claude Haiku 4.583.0%
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Estimated
~#41 of 54 (estimated)~1379 Elo
  1. #1Claude Opus 5.51501
  2. #2Claude Fable 5.1Few votes1496
  3. #40Qwen3 Next 80B A3B Instruct1386
  4. ~#41DiffusionGemma 26B A4BEstimated1379
  5. #41GLM-4.6VFew votes1361
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#41 of 54 (estimated)~1397 Elo
  1. #1Claude Opus 5.51509
  2. #2Claude Fable 5.11501
  3. #40Qwen3 Next 80B A3B Instruct1399
  4. ~#41DiffusionGemma 26B A4BEstimated1397
  5. #41GLM-4.6VFew votes1378
No published score — estimated from related benchmarks

Vision

MMMUEstimated
~#2 of 3 (estimated)~71.3%
  1. #1Claude Haiku 4.573.2%
  2. ~#2DiffusionGemma 26B A4BEstimated71.3%
  3. #2Llama 4 Maverick69.4%
No published score — estimated from related benchmarks
CharXiv ReasoningEstimated
~last of 9 (estimated)below measured range (<74.5%)
  1. #1Gemini 3.8 Flash86.2%
  2. #2Kimi K384.8%
  3. #8Gemini 3.5 Flash-Lite74.5%
  4. ~#9DiffusionGemma 26B A4BEstimated<74.5%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#5 of 8 (estimated)~80.5%
  1. #1MiniMax-M391.6%
  2. #2Kimi K391.1%
  3. #4Gemini 3.1 Pro85.3%
  4. ~#5DiffusionGemma 26B A4BEstimated80.5%
  5. #5Claude Haiku 4.579.6%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#31 of 43 (estimated)~8.0%
  1. #1GPT-6 Astra31.0%
  2. #2Muse Spark 1.326.6%
  3. #30Ling-3.0-flash-VL8.4%
  4. ~#31DiffusionGemma 26B A4BEstimated8.0%
  5. #31GPT-5.4 nano7.8%
No published score — estimated from related benchmarks
ChartographyEstimated
~last of 23 (estimated)below measured range (<9.0%)
  1. #1GPT-6 Astra71.0%
  2. #2Claude Opus 5.566.3%
  3. #22Mistral Large 39.0%
  4. ~#23DiffusionGemma 26B A4BEstimated<9.0%
No published score — estimated from related benchmarks

Hard reasoning

AIME (latest)Estimated
~#14 of 25 (estimated)~62.9%
  1. #1Inkling97.1%
  2. #2Muse Glimmer94.7%
  3. #13Phi-4-reasoning-plus78.0%
  4. ~#14DiffusionGemma 26B A4BEstimated62.9%
  5. #14Mistral Large 338.0%
No published score — estimated from related benchmarks