Google DeepMindopen-weight
Released: 2026-04-02Artificial Analysis

Gemma 4 26B A4B

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...

Intelligence Index
—/ 100
Artificial Analysis Intelligence Index (independent composite)
Throughput Speed
71tok/sec
Output streaming throughput
Time to First Answer
26s
Time to first answer token, including any thinking
Cost to Run the Index
—USD
Artificial Analysis cost of the full Index run

Measured results · Artificial Analysis

Independent evaluations of Gemma 4 26B A4B (Reasoning configuration). Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.

Intelligence Index
—
Cost to run the Index
—
Time per task
—
Omniscience index
—
Reasoning model
Yes
Cached-input price
$0.05 / 1M

Reasoning & knowledge

Coding & terminal agents

Agentic work

Long context & documents

Multimodal & science

Reasoning-effort variants evaluated

VariantIntelligence IndexCost to run IndexOutput speed (tok/s)
Reasoning——71.4
Non-reasoning——63.0
Full analysis on Artificial Analysis

Sourced Benchmark Evaluations (14)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
GPQANearing Saturation79.2%
78
2026-11-15independentSource ↗
Humanity's Last ExamActive19.3%
93
2026-11-15independentSource ↗
MMMU-ProActive69.2%
93
2026-11-15independentSource ↗
IFBenchActive72.4%
93
2026-11-15independentSource ↗
SciCodeActive40%
93
2026-11-15independentSource ↗
τ²-Bench TelecomActive43.6%
93
2026-11-15independentSource ↗
τ³-BankingActive12%
93
2026-11-15independentSource ↗
Terminal-Bench HardActive13.6%
93
2026-11-15independentSource ↗
Terminal-Bench 2.1Active39%
93
2026-11-15independentSource ↗
AA-LCR (Long Context Reasoning)Active65.7%
93
2026-11-15independentSource ↗
CritPtActive0%
93
2026-11-15independentSource ↗
HLE-DiamondActive6.2%
93
2026-11-15independentSource ↗
AutomationBench-AAActive1.9%
93
2026-11-15independentSource ↗
ITBench-AAActive23.6%
93
2026-11-15independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run google-gemma-4-26b-a4b

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google-gemma-4-26b-a4b",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for Gemma 4 26B A4B represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.