Compare · ModelsLive · 2 picked · head to head
Gemma 3 27B vs Claude 3.5 Haiku
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemma 3 27B wins on 5/6 benchmarks
Gemma 3 27B wins 5 of 6 shared benchmarks. Leads in knowledge · math.
Category leads
coding·Claude 3.5 Haikuknowledge·Gemma 3 27Bmath·Gemma 3 27B
Hype vs Reality
Attention vs performance
Gemma 3 27B
#130 by perf·no signal
Claude 3.5 Haiku
#159 by perf·no signal
Best value
Gemma 3 27B
22.7x better value than Claude 3.5 Haiku
Gemma 3 27B
351.7 pts/$
$0.12/M
Claude 3.5 Haiku
15.5 pts/$
$2.40/M
Vendor risk
Who is behind the model
Google DeepMind
$4.00T·Tier 1
Anthropic
$380.0B·Tier 1
Head to head
6 benchmarks · 2 models
Gemma 3 27BClaude 3.5 Haiku
Aider polyglot
Claude 3.5 Haiku leads by +23.1
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
Claude 3.5 Haiku
28.0
GeoBench
Gemma 3 27B leads by +18.0
GeoBench · tests geographic knowledge and spatial reasoning across countries, landmarks, coordinates, and geopolitical understanding.
Gemma 3 27B
52.0
Claude 3.5 Haiku
34.0
GPQA diamond
Gemma 3 27B leads by +14.3
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
31.8
Claude 3.5 Haiku
17.5
Lech Mazur Writing
Gemma 3 27B leads by +6.4
Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication.
Gemma 3 27B
79.9
Claude 3.5 Haiku
73.5
MATH level 5
Gemma 3 27B leads by +27.7
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
Claude 3.5 Haiku
46.4
OTIS Mock AIME 2024-2025
Gemma 3 27B leads by +15.4
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemma 3 27B
19.6
Claude 3.5 Haiku
4.2
Full benchmark table
| Benchmark | Gemma 3 27B | Claude 3.5 Haiku |
|---|---|---|
Aider polyglot Aider Polyglot · measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework. | 4.9 | 28.0 |
GeoBench GeoBench · tests geographic knowledge and spatial reasoning across countries, landmarks, coordinates, and geopolitical understanding. | 52.0 | 34.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 31.8 | 17.5 |
Lech Mazur Writing Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication. | 79.9 | 73.5 |
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 | 46.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 19.6 | 4.2 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.08 | $0.16 | 131K tokens (~66 books) | $1.00 | |
| $0.80 | $4.00 | 200K tokens (~100 books) | $16.00 |
People also compared