# Google Gemma 3 4B

As of October 22, 2026, Google Gemma 3 4B ranks #120 of 123 models overall with 22.6 ±13.5 on the LLMBenchmark Index. Gemma 3 4B ranks #101 of 108 in math, its strongest category. Google Gemma 3 4B costs $0.045 per million input tokens and $0.075 per million output tokens. Gemma 3 4B was released on March 13, 2025.

Page: https://llmbenchmark.ai/models/gemma-3-4b-it
Data from October 22, 2026. Every figure below names the board it was read from and the day.

- Maker: Google
- Released: March 13, 2025
- Price per million input tokens: $0.045
- Price per million output tokens: $0.075
- Context window: 131K tokens
- Overall rank: 120 of 123, by the LLMBenchmark Index
- LLMBenchmark Index: 22.6 ±13.5 (90% margin, from 4 of 7 inputs)
- Artificial Analysis Intelligence Index: 4.8

Benchmark results (42 of 195):

- AA Intelligence Index: 4.8 (source: Artificial Analysis (via OpenRouter), read October 22, 2026)
- AA Coding Index: 2.7 (source: Artificial Analysis (via OpenRouter), read October 22, 2026)
- AA Math Index: 12.7 (source: Artificial Analysis, read October 22, 2026)
- GPQA Diamond (Artificial Analysis run): 29.1% (source: Artificial Analysis, read October 22, 2026)
- Humanity's Last Exam (Artificial Analysis run): 5.3% (source: Artificial Analysis, read October 22, 2026)
- MMLU Pro (Artificial Analysis run): 41.7% (source: Artificial Analysis, read October 22, 2026)
- LiveCodeBench (Artificial Analysis run): 11.2% (source: Artificial Analysis, read October 22, 2026)
- Terminal-Bench Hard (Artificial Analysis run): 0.8% (source: Artificial Analysis, read October 22, 2026)
- IFBench: 28.3% (source: Artificial Analysis, read October 22, 2026)
- Long context reasoning (AA-LCR): 6.7% (source: Artificial Analysis, read October 22, 2026)
- τ²-Bench Telecom: 5.0% (source: Artificial Analysis, read October 22, 2026)
- AIME 2025 (Artificial Analysis run): 12.7% (source: Artificial Analysis, read October 22, 2026)
- GPQA Diamond (Epoch AI run): 23.2% (source: Epoch AI benchmarking hub, read October 22, 2026)
- OTIS Mock AIME: 7.5% (source: Epoch AI benchmarking hub, read October 22, 2026)
- DTBench: 50.9% (source: Epoch AI benchmarking hub, read October 22, 2026)
- LMCA: 2.8% (source: Epoch AI benchmarking hub, read October 22, 2026)
- Epoch Capabilities Index: 115.9 (source: Epoch AI benchmarking hub, read October 22, 2026)
- LMArena Text: 1291 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: coding: 1230 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: hard prompts: 1253 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: math: 1239 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: instruction following: 1239 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: creative writing: 1270 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: English: 1300 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: non-English: 1273 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: German: 1281 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: Russian: 1294 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: multi-turn: 1255 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: longer queries: 1273 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: expert: 1223 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: hard prompts in English: 1253 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: software and IT: 1269 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: mathematical work: 1256 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: science: 1306 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: medicine and healthcare: 1342 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: legal and government: 1327 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: business and finance: 1311 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: writing and language: 1288 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- LMArena Text: entertainment and media: 1233 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)
- Berkeley Function Calling Leaderboard (BFCL): 19.6% (source: Berkeley Function Calling Leaderboard, read October 22, 2026)
- Kagi LLM Benchmark: 25.2% (source: Kagi LLM Benchmarking Project, read October 22, 2026)
- SciCode: 0.0% (source: Kaggle Benchmarks, read October 22, 2026)