Across 3 comparable benchmarks, Gemma 3 - 27B (IT) has better reported results on 3 and worse on 0. 2 further shared results are listed below, each with the reason it is not counted.
Moderate evidence3 of 5 shared results countedHow we compare
Price basis: Listed rates vs standard rates. The price section below shows the applicable periods.
| Fact | LLaMA2 70B | Gemma 3 - 27B (IT) |
|---|---|---|
| Context window | 4K tokens | 128K tokens |
| Input / 1M | — | $0.080 |
| Output / 1M | — | $0.160 |
| Blended (4:1) | — | $0.096 |
| Status | Available | Available |
| Released | July 18, 2023 | March 12, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 1 shared benchmark grouped by capability on a 0–100 scale. How averages are built
Gemma 3 - 27B (IT) ahead on 1 of 1 counted
No shared comparable results for Coding, Abstract reasoning, Science & health, Computer & tool use, Research & search, Long context, Multimodal, Professional work & productivity, Cybersecurity.
Published scores, with matching modes where available. Gaps show LLaMA2 70B relative to Gemma 3 - 27B (IT). Reading the results
| Benchmark | LLaMA2 70B | Gemma 3 - 27B (IT) | Gap |
|---|---|---|---|
| HumanEvalCode Generation & Editing · AccuracyNot counted · Different or unrecorded test environments — excluded from the tally | 87.0Historical · pass@100· pass-at-100 | 87.8≠Normal · no tools· legacy-default | —#34 vs #32 of 100 |
| MBPPCode Generation & Editing · Pass rate (%)Not counted · Different or unrecorded test environments — excluded from the tally | 83.1Historical · pass@100· pass-at-100 | 74.4≠Normal · no tools· legacy-default | —#15 vs #25 of 67 |
| GSM8KMathematics · Accuracy | 56.8Historical | 95.9≠Normal · no tools | −39.1 pts#39 vs #6 of 67 |
| MMLUKnowledge Exams · Accuracy | 68.9Historical | 78.6≠Normal · no tools | −9.7 pts#69 vs #51 of 114 |
| MMLU-ProKnowledge Exams · Accuracy | 37.5Reported best | 67.5≠Normal · no tools | −30.0 pts#137 vs #99 of 153 |
Code Generation & Editing · Accuracy
Not counted · Different or unrecorded test environments — excluded from the tally
Code Generation & Editing · Pass rate (%)
Not counted · Different or unrecorded test environments — excluded from the tally
Mathematics · Accuracy
A dash means no result is available.
| Benchmark | LLaMA2 70B | Gemma 3 - 27B (IT) |
|---|---|---|
| AGIEval | 54.2 | — |
| AIME 2024 | — | 25.3 |
| GPQA | — | 36.8 |
| GPQA Diamond | — | 42.4 |
| MATH | — | 89.0 |
| Aider-Polyglot | — | 4.9 |
| LiveCodeBench | — | 29.7 |
| ECI | 114 | — |
| Creative Writing | — | 1266 |
| DROP | — | 77.2 |
| MMMU | — | 64.9 |
| BBH | — | 87.6 |
| HellaSwag | — | 85.6 |
| SimpleQA | — | 10.0 |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
LLaMA2 70B: Listed text rates; pricing tier not confirmed.
Gemma 3 - 27B (IT): Standard pricing. DeepInfra official models API (google/gemma-3-27b-it) verified 2026-11-13
| Price / 1M tokens | LLaMA2 70B | Gemma 3 - 27B (IT) | Ratio |
|---|---|---|---|
| Input | — | $0.08 | — |
| Output | — | $0.16 | — |
| Cached input | — | — | — |
| Cache write | — | — | — |
| Blended | — | $0.096 | — |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
| Attribute | LLaMA2 70B | Gemma 3 - 27B (IT) |
|---|---|---|
| Availability | Available | Available |
| Provider | Facebook AI Research Lab | Google Deep Mind |
| Model type | Foundation model | Chat model |
| First release | July 18, 2023 | March 12, 2025 |
| Context window | 4K tokens | 128K tokens |
| Total parameters | 70B | 27B |
| Architecture | Dense | Dense |
| Documented capabilities | — | Multilingual |
| Input modalities | — | Text, Image |
| Output modalities | — | Text |
| Weights | Open weights | Open weights |
| Licence | Llama 2 Community License Agreement | Gemma Terms of Use |
| Commercial use | Commercial use allowed | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used LLaMA2 70B or Gemma 3 - 27B (IT)? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.