GoogleGemma 2 27B
Gemma 2 27B is a Google open-weights model, released 24 Jun 2024. Its best configuration ranks #643 of 785 on the BenchLeader Index at 41.0 ±4.6, in the lower half. The ± is the point: 286 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (44) and lowest in reasoning (33). At $0.650 per million tokens blended it is mid-priced. Output speed of 36 tokens per second puts it in the slowest quarter, with a first token in 1.3 s. Last measured 22 Oct 2026.
- Blended price
- $0.650/M
- $0.650 in · $0.650 out
- Output speed
- 36 tok/s
- OpenRouter traffic, 7-day median; not yet measured by Artificial Analysis
- First answer
- 1.27 s
- Context
- 8k
- Full answer
- –
- Cost per run
- –
- Released
- 24 Jun 2024
- Overall index41
- Reasoning33
- Coding41
- Maths36
- Knowledge34
- Human preference44
Versions
Google has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Gemma 3 27B | 12 Mar 2025 | 38.9 | #698 |
| Gemma 2 27Bthis page | 24 Jun 2024 | 41.0 | #643 |
How its index has moved
8 Sept 2026 to 19 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| GPQA Diamond | 36.5%#256 | Epoch AI Benchmarking Hub | |
| Arena Hard Prompts | 1282#271 | Arena (LMArena) | |
| LMCAnot in index | 7.1%#185 | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | 48.0%#220 | Epoch AI Benchmarking Hub |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Coding | 1304#277 | Arena (LMArena) |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| OTIS Mock AIME | 1.4%#265 | Epoch AI Benchmarking Hub | |
| MATH Level 5 | 27.9%#69 | Epoch AI Benchmarking Hub | |
| Arena Maths | 1248#268 | Arena (LMArena) |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| LegalBench | 68.0%#127 | Vals AI | |
| MMLUnot in index | 75.7%#36 | Epoch AI Benchmarking Hub |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Instruction Followingnot in index | 1271#265 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1290#262 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1293#226 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1280#255 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1301#256 | Arena (LMArena) |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 122.1#167 | Epoch AI Benchmarking Hub |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| NextBit | 48 tok/s | 1.47 s | $0.650 | $0.650 | 8k | int4 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0005 | – | 9.6 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0082 | – | 17.9 s |
| Code edit | 6,000 / 1,500 | $0.0049 | – | 42.9 s |
| Agentic coding session | 60,000 / 4,000 | $0.042 | – | 1.9 min |
| Structured extraction | 2,000 / 200 | $0.0014 | – | 6.8 s |
More from Google
See also
Questions people ask
- What are Gemma 2 27B's benchmark scores?
- As of 25 Oct 2026, Gemma 2 27B scores 41.0 ±4.6 on the BenchLeader Index, strongest in human preference (44) and weakest in reasoning (33). That rests on 8 independent benchmark results, and every score on this page links to its publisher.
- How does Gemma 2 27B rank?
- Gemma 2 27B is #643 of 785 ranked configurations on 25 Oct 2026. 286 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Gemma 2 27B cost per 1M tokens?
- As of 25 Oct 2026, Gemma 2 27B costs $0.650 per 1M input tokens and $0.650 per 1M output tokens, $0.650 blended (three input tokens per output token).
- When was Gemma 2 27B released?
- Google released Gemma 2 27B on 24 Jun 2024.
Data as of 25 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Gemma 2 27B: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/gemma-2-27b.