Google DeepMindopen-weight
Released: 2025-04-12

Gemma 3 27B Preview

Third-generation Gemma flagship with extended 128K context, native multimodal visual reasoning, and enhanced code comprehension.

Intelligence Index
84.2/ 100
Calibrated multi-domain composite
Throughput Speed
88tok/sec
Output streaming throughput
Time To First Token
320ms
Initial chunk server latency
LiveBench Contam-Free
76.2%
September 2026 suite score

LiveBench Multi-Domain Evaluation Breakdown

Monthly refreshed contamination-free test sets
Logical Reasoning80.5%
Coding & Repo Repair75.8%
Mathematics (AIME)79.8%
Data Analysis & Tables74.2%
Language & Comprehension73.5%
Instruction Following74.0%

Sourced Benchmark Evaluations (3)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-2Nearing Saturation65.5%
75
2025-05-10independentSource ↗
SWE-bench VerifiedSaturated60%
25
2025-04-21independentSource ↗
IFEvalDeprecated74%
25
2025-04-16independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run gemma-3-27b

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma-3-27b",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for Gemma 3 27B Preview 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.