Gemini 3.7 Flash vs Step 5 Preview
Updated October 21, 2026. 56 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Gemini 3.7 Flash and Step 5 Preview are level for coding, 69.9 against 71.4 on the Vibe Coding Index, so price and context decide.
Pick Gemini 3.7 Flash when what matters is LMArena WebDev or IOI (Vals AI), where it leads. Pick Step 5 Preview when output tokens cost 1.4x less, or when what matters is LiveBench coding or Terminal-Bench 4 (Vals AI), where it leads.
- Shared results
- 56
- Gemini 3.7 Flash only
- 56
- Step 5 Preview only
- 5
- Like-for-like categories
- 8 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
No call
Level: 69.9 against 71.4, inside 2 points.
2 benchmarks behind each score
Intelligence
AA Intelligence Index
Step 5 Preview
Step 5 Preview leads, 43.7 against 39.6.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Gemini 3.7 Flash has a GPQA Diamond result, 93.9%.
Directional only
Knowledge
MMLU Pro
No call
Only Gemini 3.7 Flash has a MMLU Pro result, 90.1%.
Directional only
Web design
Design Arena: websites
No call
Only Gemini 3.7 Flash has a Design Arena: websites result, 1304.
Directional only
Long documents
Listed context window
No call
Both take 1.05M tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Step 5 Preview
Gemini 3.7 Flash $0.0026, Step 5 Preview $0.0024 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Gemini 3.7 Flash has the number, $1.78.
Directional only
At a glance
- Coding index
- Gemini 3.7 Flash 69.9 from 2 benchmarks, Step 5 Preview 71.4 from 2 benchmarks.
- Widest category
- Math, 10.7 points apart on average over 3 shared benchmarks.
- Widest single gap
- LiveBench browser: Gemini 3.7 Flash 56.0% against Step 5 Preview 24.0%, 32.0 points apart on a 0 to 100 scale.
- Output price
- Gemini 3.7 Flash $3.75 against Step 5 Preview $2.70 per million output tokens, 1.4x apart.
- Context window
- Both take 1.05M tokens.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Gemini 3.7 Flash | Bars | Step 5 Preview | Ties |
|---|---|---|---|---|
| Coding | 8 | 4 | ||
| Intelligence | 0 | 1 | ||
| Agentic work | 7 | 2 | ||
| Reasoning | 7 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| Web design | 0 | 1 | ||
| General | 18 | 1 | ||
| All shared boards | 47 | 9 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index84
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench coding | 80.5% | 85.6% | Step 5 Preview |
| LMArena WebDev | 1591 | 1567 | Gemini 3.7 Flash |
| AA Coding Index | 71.5 | Not measured | Only Gemini 3.7 Flash measured |
| LiveCodeBench (Vals AI) | 88.7% | Not measured | Only Gemini 3.7 Flash measured |
| SWE-bench Verified (Vals AI) | 80.8% | Not measured | Only Gemini 3.7 Flash measured |
| Terminal-Bench 4 (Vals AI) | 12.1% | 31.8% | Step 5 Preview |
| Vibe Code Bench (Vals AI) | 70.4% | 70.4% | Step 5 Preview |
| Code migration (Vals AI) | 34.8% | 39.0% | Step 5 Preview |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Gemini 3.7 Flash measured |
| IOI (Vals AI) | 67.8% | 38.1% | Gemini 3.7 Flash |
| Terminal-Bench 2.1 (Vals AI) | 77.5% | Not measured | Only Gemini 3.7 Flash measured |
| SWE-bench Multimodal (Vals AI) | Not measured | 40.8% | Only Step 5 Preview measured |
| SkillsBench (Vals AI) | 65.9% | Not measured | Only Gemini 3.7 Flash measured |
| CyberBench (Vals AI) | 43.8% | Not measured | Only Gemini 3.7 Flash measured |
| SciCode | 59.8% | 58.9% | Gemini 3.7 Flash |
| FrontierCode | 43.6% | Not measured | Only Gemini 3.7 Flash measured |
| FrontierSWE | 20.3% | Not measured | Only Gemini 3.7 Flash measured |
| ALE-Bench | 904.3 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: coding | 1499 | 1485 | Gemini 3.7 Flash |
| LMArena Text: software and IT | 1498 | 1473 | Gemini 3.7 Flash |
| LMArena WebDev: React | 1587 | 1583 | Gemini 3.7 Flash |
| LMArena WebDev: HTML | 1593 | 1556 | Gemini 3.7 Flash |
| LMArena WebDev: from an image | 1599 | 1550 | Gemini 3.7 Flash |
| Neither model has a result on the other 15 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 39.6 | 43.7 | Step 5 Preview |
| Epoch Capabilities Index | 158.0 | Not measured | Only Gemini 3.7 Flash measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index72
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1224 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: agents building full-stack apps | 1208 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: agents building mobile apps | 1230 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: agents building Android apps | 1247 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: agents building Godot games | 1264 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: agents building games | 1235 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: spreadsheets | 1028 | Not measured | Only Gemini 3.7 Flash measured |
| LiveBench research | 76.5% | 72.8% | Gemini 3.7 Flash |
| LiveBench browser | 56.0% | 24.0% | Gemini 3.7 Flash |
| LiveBench long agentic | 28.3% | 27.5% | Gemini 3.7 Flash |
| SREBench (Vals AI) | 4.6% | 0.4% | Gemini 3.7 Flash |
| Finance Agent (Vals AI) | 59.0% | 50.7% | Gemini 3.7 Flash |
| Tax Agent Bench (Vals AI) | 22.8% | 27.3% | Step 5 Preview |
| Legal Research Bench (Vals AI) | 34.6% | 40.4% | Step 5 Preview |
| Excel Modeling Benchmark (Vals AI) | 71.3% | 57.1% | Gemini 3.7 Flash |
| BioMysteryBench (Vals AI) | Not measured | 67.4% | Only Step 5 Preview measured |
| Terminal-Bench-Science (Vals AI) | 7.1% | 2.9% | Gemini 3.7 Flash |
| Vals Science Index | Not measured | 29.6% | Only Step 5 Preview measured |
| APEX-Agents | 67.8% | Not measured | Only Gemini 3.7 Flash measured |
| Remote Labor Index | 5.0% | Not measured | Only Gemini 3.7 Flash measured |
| Neither model has a result on the other 19 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index70
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| GPQA Diamond | 93.9% | Not measured | Only Gemini 3.7 Flash measured |
| LiveBench reasoning | 87.8% | 83.8% | Gemini 3.7 Flash |
| MysteryMechanism (Vals AI) | Not measured | 18.5% | Only Step 5 Preview measured |
| MMMU Pro (Vals AI) | 89.0% | Not measured | Only Gemini 3.7 Flash measured |
| GPQA Diamond (Artificial Analysis run) | 94.5% | Not measured | Only Gemini 3.7 Flash measured |
| Humanity's Last Exam (Artificial Analysis run) | 47.9% | 46.5% | Gemini 3.7 Flash |
| GPQA Diamond (Epoch AI run) | 94.8% | Not measured | Only Gemini 3.7 Flash measured |
| ARC-AGI | 95.5% | Not measured | Only Gemini 3.7 Flash measured |
| ARC-AGI-2 | 84.6% | Not measured | Only Gemini 3.7 Flash measured |
| Chess puzzles | 66.0% | Not measured | Only Gemini 3.7 Flash measured |
| Mystery game puzzles | 37.0% | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: hard prompts | 1495 | 1461 | Gemini 3.7 Flash |
| LMArena Text: expert | 1512 | 1485 | Gemini 3.7 Flash |
| LMArena Text: hard prompts in English | 1487 | 1454 | Gemini 3.7 Flash |
| LMArena Vision: diagrams | 1329 | 1308 | Gemini 3.7 Flash |
| LMArena Vision: homework | 1320 | 1265 | Gemini 3.7 Flash |
| Neither model has a result on the other 5 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| OTIS Mock AIME | 97.2% | Not measured | Only Gemini 3.7 Flash measured |
| FrontierMath, tiers 1 to 3 | 71.6% | Not measured | Only Gemini 3.7 Flash measured |
| FrontierMath, tier 4 | 36.6% | Not measured | Only Gemini 3.7 Flash measured |
| ProofBench | 58.0% | 42.0% | Gemini 3.7 Flash |
| LMArena Text: math | 1511 | 1458 | Gemini 3.7 Flash |
| LMArena Text: mathematical work | 1502 | 1460 | Gemini 3.7 Flash |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| MMLU Pro | 90.1% | Not measured | Only Gemini 3.7 Flash measured |
| LegalBench (Vals AI) | 87.3% | Not measured | Only Gemini 3.7 Flash measured |
| TaxEval (Vals AI) | 74.7% | Not measured | Only Gemini 3.7 Flash measured |
| MortgageTax (Vals AI) | 66.7% | Not measured | Only Gemini 3.7 Flash measured |
| MedCode (Vals AI) | 53.4% | Not measured | Only Gemini 3.7 Flash measured |
| SimpleQA Verified | 69.2% | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: science | 1511 | 1454 | Gemini 3.7 Flash |
| LMArena Text: medicine and healthcare | 1494 | 1439 | Gemini 3.7 Flash |
| LMArena Text: legal and government | 1496 | 1437 | Gemini 3.7 Flash |
| LMArena Text: business and finance | 1479 | 1435 | Gemini 3.7 Flash |
| Neither model has a result on the other 7 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| Design Arena: websites | 1304 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: UI components | 1299 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: game dev | 1318 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: data viz | 1315 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: 3D | 1308 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: all code categories | 1308 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: SVG | 1283 | Not measured | Only Gemini 3.7 Flash measured |
| Design Arena: ASCII art | 1276 | Not measured | Only Gemini 3.7 Flash measured |
| LiveBench design | 73.2% | 76.9% | Step 5 Preview |
GeneralLike-for-likePublic boards, shown as published, kept out of the index181
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench data analysis | 68.0% | 63.5% | Gemini 3.7 Flash |
| Public Benefits Bench (Vals AI) | Not measured | 58.1% | Only Step 5 Preview measured |
| SAGE (Vals AI) | 49.2% | Not measured | Only Gemini 3.7 Flash measured |
| MedScribe (Vals AI) | 83.9% | Not measured | Only Gemini 3.7 Flash measured |
| Long context reasoning (AA-LCR) | 83.0% | 88.3% | Step 5 Preview |
| DTBench | 96.8% | Not measured | Only Gemini 3.7 Flash measured |
| LMCA | 50.4% | Not measured | Only Gemini 3.7 Flash measured |
| Furniture assembly | 26.7% | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text | 1489 | 1440 | Gemini 3.7 Flash |
| LMArena Text: instruction following | 1484 | 1438 | Gemini 3.7 Flash |
| LMArena Text: creative writing | 1488 | 1409 | Gemini 3.7 Flash |
| LMArena Vision | 1314 | 1265 | Gemini 3.7 Flash |
| LMArena Text: English | 1483 | 1447 | Gemini 3.7 Flash |
| LMArena Text: non-English | 1487 | 1428 | Gemini 3.7 Flash |
| LMArena Text: Chinese | 1552 | 1500 | Gemini 3.7 Flash |
| LMArena Text: German | 1506 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: Japanese | 1511 | 1383 | Gemini 3.7 Flash |
| LMArena Text: Korean | 1481 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: French | 1500 | 1438 | Gemini 3.7 Flash |
| LMArena Text: Spanish | 1495 | 1441 | Gemini 3.7 Flash |
| LMArena Text: Russian | 1515 | 1433 | Gemini 3.7 Flash |
| LMArena Text: Polish | 1503 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Text: multi-turn | 1485 | 1437 | Gemini 3.7 Flash |
| LMArena Text: longer queries | 1492 | 1449 | Gemini 3.7 Flash |
| LMArena Text: writing and language | 1491 | 1425 | Gemini 3.7 Flash |
| LMArena Text: entertainment and media | 1477 | 1412 | Gemini 3.7 Flash |
| LMArena Vision: English | 1309 | 1290 | Gemini 3.7 Flash |
| LMArena Vision: Chinese | 1394 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Vision: reading text | 1322 | 1282 | Gemini 3.7 Flash |
| LMArena Vision: humor | 1299 | Not measured | Only Gemini 3.7 Flash measured |
| LMArena Vision: creative writing | 1312 | Not measured | Only Gemini 3.7 Flash measured |
| Neither model has a result on the other 15 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemini 3.7 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Step 5 Preview costs less
Repository review
50K input + 3K output tokens
Gemini 3.7 Flash costs less
Agent loop
200K input + 10K output tokens
Gemini 3.7 Flash costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Google Gemini 3.7 Flash | StepFun Step 5 Preview |
|---|---|---|
| Maker | StepFun | |
| Weights | No public weights listed | Open weights |
| Released | August 13, 2026 | October 8, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $0.75 | $1.00 |
| Output $/Mtok | $3.75 | $2.70 |
| Coding index | 69.9 | 71.4 |
| Independent measurements behind it | 2 | 2 |
| Cost per solved task | $1.78 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemini 3.7 Flash better than Step 5 Preview for coding?
Gemini 3.7 Flash and Step 5 Preview are level for coding, 69.9 against 71.4 on the Vibe Coding Index, so price and context decide. Pick Gemini 3.7 Flash when what matters is LMArena WebDev or IOI (Vals AI), where it leads. Pick Step 5 Preview when output tokens cost 1.4x less, or when what matters is LiveBench coding or Terminal-Bench 4 (Vals AI), where it leads.
Which is cheaper, Gemini 3.7 Flash or Step 5 Preview?
Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; Step 5 Preview costs $1.00 and $2.70. Cost per solved task is known for Gemini 3.7 Flash only.
Which has the bigger context window, Gemini 3.7 Flash or Step 5 Preview?
Neither: both take 1.05M tokens.
How many benchmarks is the Gemini 3.7 Flash vs Step 5 Preview comparison based on?
56 benchmarks have a result for both models as of October 21, 2026. Gemini 3.7 Flash's coding index rests on 2 independent benchmarks and Step 5 Preview's on 2. Where only one model has a result, the page says so instead of naming a winner.