Model comparison
Devstral Small 2505 vs Gemma 3 27B
Devstral Small 2505 is the stronger model overall, scoring 34.3 to 30.8 on the Noometry Index. Gemma 3 27B costs 1.5× less per token, which makes it the better buy when Devstral Small 2505's lead doesn't matter for your workload.
Last verified . 3 shared benchmarks.
Mistral AI
GoogleSummary
- They share 3 benchmarks with published results for both. Devstral Small 2505 scores higher in 2 categories and Gemma 3 27B in 0 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where Devstral Small 2505 leads 38.9 to 22.5.
- The biggest single-benchmark swing is SciCode: 28.8% for Devstral Small 2505 and 21.2% for Gemma 3 27B.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $0.10 / $0.30 for Devstral Small 2505.
- Gemma 3 27B accepts more context: 131K tokens versus 128K.
Side by side
| Devstral Small 2505 | Gemma 3 27B | |
|---|---|---|
| Provider | Mistral AI | |
| Noometry Index | 34.3 | 30.8 |
| Released | 2025-05-07 | 2025-03-11 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 128K | 8K |
| Input $ / M tokens | $0.10 | $0.08 |
| Output $ / M tokens | $0.30 | $0.16 |
| Results tracked | 4 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Devstral Small 2505 leads
Devstral Small 2505: 38.9 (#166), Gemma 3 27B: 22.5 (#334)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| SciCode | 28.8% | 21.2% |
| SWE-bench Verified (bash only) | 56.4% | — |
| Aider Polyglot | — | 4.9% |
| LiveBench Coding | — | 39.9% |
| LMArena Coding | — | 1322 |
Agentic & Tool Use Not comparable
Devstral Small 2505: —, Gemma 3 27B: 25.1 (#110)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 29.5% |
Reasoning Devstral Small 2505 leads
Devstral Small 2505: 19.7 (#252), Gemma 3 27B: 16.7 (#301)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| Kagi LLM Benchmark | 37.7% | 40.4% |
| CritPt | 0% | 0% |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | — | 43.8% |
| LMArena Hard Prompts | — | 1340 |
| DTBench | — | 52.5% |
| LiveBench Data Analysis | — | 51.5% |
| LMCA | — | 12.3% |
| Epoch Capabilities Index | — | 130.04 |
| LiveBench | — | 50% |
Math Not comparable
Devstral Small 2505: —, Gemma 3 27B: 25.9 (#265)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 22.5% |
| LiveBench Math | — | 55.4% |
| LMArena Math | — | 1312 |
| MATH Level 5 | — | 74% |
Knowledge Not comparable
Devstral Small 2505: —, Gemma 3 27B: 25.5 (#261)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| GPQA Diamond | — | 47.7% |
| Confabulations | — | 40.3% |
| Vectara Hallucination Rate | — | 7.4% |
| LMArena Expert | — | 1304 |
Multimodal Not comparable
Devstral Small 2505: —, Gemma 3 27B: 32.6 (#100)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| LMArena Vision | — | 1164 |
| GeoBench | — | 52% |
Multilingual Not comparable
Devstral Small 2505: —, Gemma 3 27B: 46.9 (#155)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| LMArena Non-English | — | 1334 |
| LMArena Chinese | — | 1346 |
| LMArena French | — | 1368 |
| LMArena German | — | 1362 |
| LMArena Japanese | — | 1287 |
| LMArena Korean | — | 1308 |
| LMArena Russian | — | 1349 |
| LMArena Spanish | — | 1349 |
Instruction Following Not comparable
Devstral Small 2505: —, Gemma 3 27B: 70.6 (#160)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| LiveBench Instruction Following | — | 74.9% |
| LMArena Instruction Following | — | 1321 |
Long Context Not comparable
Devstral Small 2505: —, Gemma 3 27B: 27.6 (#293)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| Fiction.LiveBench | — | 33.3% |
| LMArena Longer Query | — | 1333 |
Writing & Preference Not comparable
Devstral Small 2505: —, Gemma 3 27B: 52.5 (#168)
| Benchmark | Devstral Small 2505 | Gemma 3 27B |
|---|---|---|
| LMArena Text | — | 1358 |
| LMArena Creative Writing | — | 1346 |
| Short-Story Creative Writing | — | 79.9% |
| EQ-Bench Creative Writing | — | 1266 |
| LMArena Multi-Turn | — | 1345 |
| LiveBench Language | — | 34.6% |
Frequently asked questions
Is Devstral Small 2505 better than Gemma 3 27B?
Devstral Small 2505 is the stronger model overall, scoring 34.3 to 30.8 on the Noometry Index. Gemma 3 27B costs 1.5× less per token, which makes it the better buy when Devstral Small 2505's lead doesn't matter for your workload.
Which is cheaper, Devstral Small 2505 or Gemma 3 27B?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; Devstral Small 2505 lists at $0.10 and $0.30.
Is Devstral Small 2505 or Gemma 3 27B better for coding?
Devstral Small 2505 scores higher on coding benchmarks: 38.9 versus 22.5 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 27B does, with 131K tokens against 128K.
How many benchmarks do Devstral Small 2505 and Gemma 3 27B share?
3 benchmarks have published results for both models. Devstral Small 2505 has 4 scored results on Noometry and Gemma 3 27B has 43.