Compare · ModelsLive · 2 picked · head to head
Gemma 3 27B vs Mixtral 8x22B Instruct
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemma 3 27B wins on 2/2 benchmarks
Gemma 3 27B wins 2 of 2 shared benchmarks. Leads in knowledge · math.
Category leads
knowledge·Gemma 3 27Bmath·Gemma 3 27B
Hype vs Reality
Attention vs performance
Gemma 3 27B
#130 by perf·no signal
Mixtral 8x22B Instruct
#208 by perf·no signal
Best value
Gemma 3 27B
59.9x better value than Mixtral 8x22B Instruct
Gemma 3 27B
351.7 pts/$
$0.12/M
Mixtral 8x22B Instruct
5.9 pts/$
$4.00/M
Vendor risk
Who is behind the model
Google DeepMind
$4.00T·Tier 1
Mistral AI
$14.0B·Tier 1
Head to head
2 benchmarks · 2 models
Gemma 3 27BMixtral 8x22B Instruct
GPQA diamond
Gemma 3 27B leads by +19.7
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
Mixtral 8x22B Instruct
12.1
MATH level 5
Gemma 3 27B leads by +49.8
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
Mixtral 8x22B Instruct
24.2
Full benchmark table
| Benchmark | Gemma 3 27B | Mixtral 8x22B Instruct |
|---|---|---|
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 | 12.1 |
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 | 24.2 |
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 | |
| $2.00 | $6.00 | 66K tokens (~33 books) | $30.00 |