Skip to content
Model Pareto

Gemma 4 26B A4B

Google DeepMindOpen weightsartificialanalysis.ai ↗
Compare with nearest rivalsCompare with…
Input
$0.10/M tok
Output
$0.34/M tok
Blended
$0.16/M tok · 3:1
Speed
~89tok/sest.
Context
256Ktokens

Long-tail import from Artificial Analysis (2026-09-24); AA variant: Gemma 4 26B A4B (Reasoning). Price = AA median host price.

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.

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.

Coding

SWE-bench VerifiedEstimated
~#11 of 14 (estimated)~71.9%
  1. #1Claude Opus 596.0%
  2. #2Claude Sonnet 585.2%
  3. #10Claude Haiku 4.573.3%
  4. ~#11Gemma 4 26B A4BEstimated71.9%
  5. #11Nemotron 3 Ultra70.7%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#18 of 19 (estimated)~50.6%
  1. #1Claude Opus 5.589.9%
  2. #2Claude Sonnet 5.581.3%
  3. #17Muse Glimmer51.2%
  4. ~#18Gemma 4 26B A4BEstimated50.6%
  5. #18Nex-N2.5-Mini43.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#28 of 31 (estimated)~45.4%
  1. #1DeepSeek V4.1 Flash74.2%
  2. #2Grok 4.772.6%
  3. #27Qwen3.8 Max (0902)51.0%
  4. ~#28Gemma 4 26B A4BEstimated45.4%
  5. #28Qwen3.8 27B42.2%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Estimated
~#27 of 84 (estimated)~8.2%
  1. #1Claude Sonnet 5.563.6%
  2. #2Claude Opus 5.559.6%
  3. #26Qwen3.8 2.4T A95B11.1%
  4. ~#27Gemma 4 26B A4BEstimated8.2%
  5. #27Qwen3.8 27B5.6%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#8 of 20 (estimated)~69.6%
  1. #1Qwen3.8-Omni-Flash92.6%
  2. #2Qwen3.8 Flash-Next91.9%
  3. #7gpt-oss-20b77.7%
  4. ~#8Gemma 4 26B A4BEstimated69.6%
  5. #8Mistral Small 464.0%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~last of 15 (estimated)below measured range (<41.1)
  1. #1Claude Opus 5.566.0
  2. #2Claude Fable 5.162.2
  3. #14GPT-6 Luna41.1
  4. ~#15Gemma 4 26B A4BEstimated<41.1
No published score — estimated from related benchmarks

Writing & chat

MMMLUEstimated
~#3 of 5 (estimated)~87.0%
  1. #1Gemini 3.1 Pro92.6%
  2. #2Gemma 4 31B88.4%
  3. ~#3Gemma 4 26B A4BEstimated87.0%
  4. #3Claude Haiku 4.583.0%
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Few votesVerified
#31 of 531423 Elo
  1. #1Claude Opus 5.51501
  2. #2Claude Fable 5.1Few votes1496
  3. #30MiniMax-M31427
  4. #31Gemma 4 26B A4BFew votes1423
  5. #32Muse GlimmerFew votes1418
lmarena.ai · 2026-09-13
LMArena Text (Elo)Verified
#30 of 531438 Elo
  1. #1Claude Opus 5.51509
  2. #2Claude Fable 5.11501
  3. #29Grok 4.71439
  4. #30Gemma 4 26B A4B1438
  5. #31Qwen3.8 27B1437
lmarena.ai · 2026-09-13
EQ-Bench Creative Writing v3 (Elo)Estimated
~#13 of 26 (estimated)~1801 Elo
  1. #1GPT-6 Astra2173
  2. #2Claude Fable 5.12162
  3. #12Qwen3.8 2.4T A95B1843
  4. ~#13Gemma 4 26B A4BEstimated1801
  5. #13Muse Glimmer1798
No published score — estimated from related benchmarks

Vision

MMMUEstimated
~#2 of 3 (estimated)~71.3%
  1. #1Claude Haiku 4.573.2%
  2. ~#2Gemma 4 26B A4BEstimated71.3%
  3. #2Llama 4 Maverick69.4%
No published score — estimated from related benchmarks
CharXiv ReasoningEstimated
~#8 of 9 (estimated)~76.6%
  1. #1Gemini 3.8 Flash86.2%
  2. #2Kimi K384.8%
  3. #7Claude Sonnet 577.0%
  4. ~#8Gemma 4 26B A4BEstimated76.6%
  5. #8Gemini 3.5 Flash-Lite74.5%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#5 of 8 (estimated)~83.0%
  1. #1MiniMax-M391.6%
  2. #2Kimi K391.1%
  3. #4Gemini 3.1 Pro85.3%
  4. ~#5Gemma 4 26B A4BEstimated83.0%
  5. #5Claude Haiku 4.579.6%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#28 of 43 (estimated)~10.9%
  1. #1GPT-6 Astra31.0%
  2. #2Muse Spark 1.326.6%
  3. #26Qwen3.7 Plus12.2%
  4. ~#28Gemma 4 26B A4BEstimated10.9%
  5. #28Muse Glimmer10.0%
No published score — estimated from related benchmarks
ChartographyEstimated
~#18 of 23 (estimated)~15.0%
  1. #1GPT-6 Astra71.0%
  2. #2Claude Opus 5.566.3%
  3. #17GLM-5.3 Flash16.3%
  4. ~#18Gemma 4 26B A4BEstimated15.0%
  5. #18Grok 4.714.7%
No published score — estimated from related benchmarks

Hard reasoning

AIME (latest)Estimated
~#10 of 25 (estimated)~84.9%
  1. #1Inkling97.1%
  2. #2Muse Glimmer94.7%
  3. #9Mistral Medium 3.586.3%
  4. ~#10Gemma 4 26B A4BEstimated84.9%
  5. #10Mistral Small 484.0%
No published score — estimated from related benchmarks

Agents

OSWorld-VerifiedEstimated
~#10 of 12 (estimated)~66.4%
  1. #1Kimi K384.8%
  2. #2Qwen3.8 27B84.3%
  3. #9Nex-N2.5-Mini71.2%
  4. ~#10Gemma 4 26B A4BEstimated66.4%
  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. ~#5Gemma 4 26B A4BEstimated<47.5%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#15 of 18 (estimated)~55.1%
  1. #1Atria Dawn Preview92.5%
  2. #2GPT-6 Astra91.5%
  3. #14Inkling77.1%
  4. ~#15Gemma 4 26B A4BEstimated55.1%
  5. #15Mistral Medium 3.548.6%
No published score — estimated from related benchmarks