Compare · ModelsLive · 2 picked · head to head
Gemma 3 27B vs GPT-4.1 Nano
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemma 3 27B wins on 4/7 benchmarks
Gemma 3 27B wins 4 of 7 shared benchmarks. Leads in general · knowledge · math.
Category leads
coding·GPT-4.1 Nanogeneral·Gemma 3 27Bknowledge·Gemma 3 27Bmath·Gemma 3 27B
Hype vs Reality
Attention vs performance
Gemma 3 27B
#212 by perf·#7 by attention
GPT-4.1 Nano
#246 by perf·no signal
Best value
Gemma 3 27B
1.1x better value than GPT-4.1 Nano
Gemma 3 27B
137.7 pts/$
$0.27/M
GPT-4.1 Nano
124.0 pts/$
$0.25/M
Vendor risk
Who is behind the model
Google DeepMind
$4.20T·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
7 benchmarks · 2 models
Gemma 3 27BGPT-4.1 Nano
Aider polyglot
GPT-4.1 Nano leads by +4.0
Aider Polyglot · measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework.
Gemma 3 27B
4.9
GPT-4.1 Nano
8.9
Dtbench
Gemma 3 27B
20.9
GPT-4.1 Nano
20.9
Fiction.LiveBench
Gemma 3 27B leads by +8.3
Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination.
Gemma 3 27B
33.3
GPT-4.1 Nano
25.0
GPQA diamond
GPT-4.1 Nano leads by +1.6
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemma 3 27B
30.3
GPT-4.1 Nano
31.9
Lmca
Gemma 3 27B leads by +8.0
Gemma 3 27B
14.4
GPT-4.1 Nano
6.4
MATH level 5
Gemma 3 27B leads by +4.0
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
Gemma 3 27B
74.0
GPT-4.1 Nano
70.0
OTIS Mock AIME 2024-2025
GPT-4.1 Nano leads by +6.4
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemma 3 27B
22.4
GPT-4.1 Nano
28.8
Full benchmark table
| Benchmark | Gemma 3 27B | GPT-4.1 Nano |
|---|---|---|
Aider polyglot Aider Polyglot · measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework. | 4.9 | 8.9 |
Dtbench | 20.9 | 20.9 |
Fiction.LiveBench Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination. | 33.3 | 25.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 30.3 | 31.9 |
Lmca | 14.4 | 6.4 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 74.0 | 70.0 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 22.4 | 28.8 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.08 | $0.45 | 131K tokens (~66 books) | $1.72 | |
| $0.10 | $0.40 | 1.0M tokens (~524 books) | $1.75 |
People also compared