Gemini 3.8 Flash vs Step 5 Preview
Gemini 3.8 Flash leads Step 5 Preview on coding, 61.6 to 61.1 on the Vibe Coding Index, and Step 5 Preview costs 1.4x less per output token ($2.70 vs $3.75 per million). Updated October 22, 2026; 61 benchmarks have a result for both.
Gemini 3.8 Flash or Step 5 Preview: which to pick
Gemini 3.8 Flash and Step 5 Preview are level for coding, 61.6 against 61.1 on the Vibe Coding Index, so price and context decide.
Pick Step 5 Preview for output tokens 1.4x cheaper ($2.70 vs $3.75 per million) and work measured by LiveBench coding or Terminal-Bench 4 (Vals AI), where it leads. Pick Gemini 3.8 Flash for work measured by LMArena WebDev or Vibe Code Bench (Vals AI), where it leads.
- Shared results
- 61
- Gemini 3.8 Flash only
- 67
- Step 5 Preview only
- 0
- Like-for-like categories
- 8 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
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, six public coding benchmarks
No call
Level: 61.6 against 61.1, inside 2 points.
6 and 4 benchmarks behind the scores
Intelligence
AA Intelligence Index
Step 5 Preview
Step 5 Preview leads, 43.7 against 40.9.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Gemini 3.8 Flash has a AA Agentic Index result, 40.2.
Directional only
Reasoning
GPQA Diamond
No call
Only Gemini 3.8 Flash has a GPQA Diamond result, 94.4%.
Directional only
Knowledge
MMLU Pro
No call
Only Gemini 3.8 Flash has a MMLU Pro result, 90.2%.
Directional only
Web design
Design Arena: websites
No call
Only Gemini 3.8 Flash has a Design Arena: websites result, 1300.
Directional only
General
Vals Index
No call
Only Gemini 3.8 Flash has a Vals Index result, 56.0%.
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 3.8 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 3.8 Flash has the number, $2.74.
Directional only
At a glance
- Coding index
- Gemini 3.8 Flash 61.6 from 6 benchmarks, Step 5 Preview 61.1 from 4 benchmarks.
- Widest category
- Knowledge, 10.1 points apart on average over 4 shared benchmarks.
- Widest single gap
- LiveBench browser: Gemini 3.8 Flash 64.0% against Step 5 Preview 24.0%, 40.0 points apart on a 0 to 100 scale.
- Output price
- Gemini 3.8 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 3.8 Flash | Bars | Step 5 Preview | Ties |
|---|---|---|---|---|
| Coding | 7 | 5 | 1 tie | |
| Intelligence | 0 | 1 | ||
| Agentic work | 9 | 2 | ||
| Reasoning | 8 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| Web design | 0 | 1 | ||
| General | 18 | 2 | ||
| All shared boards | 49 | 11 | 1 tie |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index75
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench coding | 82.2% | 85.6% | Step 5 Preview |
| LMArena WebDev | 1585 | 1567 | Gemini 3.8 Flash |
| AA Coding Index | 76.3 | Not measured | Only Gemini 3.8 Flash measured |
| LiveCodeBench (Vals AI) | 89.5% | Not measured | Only Gemini 3.8 Flash measured |
| SWE-bench Verified (Vals AI) | 80.0% | Not measured | Only Gemini 3.8 Flash measured |
| Terminal-Bench 4 (Vals AI) | 19.7% | 31.8% | Step 5 Preview |
| Vibe Code Bench (Vals AI) | 78.7% | 70.4% | Gemini 3.8 Flash |
| Code migration (Vals AI) | 36.5% | 39.0% | Step 5 Preview |
| ProgramBench (Vals AI) | 1.0% | Not measured | Only Gemini 3.8 Flash measured |
| IOI (Vals AI) | 56.9% | 38.1% | Gemini 3.8 Flash |
| Terminal-Bench 2.1 (Vals AI) | 81.3% | Not measured | Only Gemini 3.8 Flash measured |
| SWE-bench Multimodal (Vals AI) | 40.8% | 40.8% | Tie |
| Vibe Code Bench 1-100 (Vals AI) | 18.8% | Not measured | Only Gemini 3.8 Flash measured |
| SkillsBench (Vals AI) | 58.0% | Not measured | Only Gemini 3.8 Flash measured |
| CyberBench (Vals AI) | 43.8% | Not measured | Only Gemini 3.8 Flash measured |
| SciCode | 56.6% | 58.9% | Step 5 Preview |
| FrontierCode | 41.2% | Not measured | Only Gemini 3.8 Flash measured |
| FrontierSWE | 19.6% | Not measured | Only Gemini 3.8 Flash measured |
| WeirdML | 84.8% | Not measured | Only Gemini 3.8 Flash measured |
| ALE-Bench | 1269.7 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: coding | 1508 | 1485 | Gemini 3.8 Flash |
| LMArena Text: software and IT | 1507 | 1473 | Gemini 3.8 Flash |
| LMArena WebDev: React | 1577 | 1583 | Step 5 Preview |
| LMArena WebDev: HTML | 1616 | 1556 | Gemini 3.8 Flash |
| LMArena WebDev: from an image | 1593 | 1550 | Gemini 3.8 Flash |
| SciCode | 12.3% | Not measured | Only Gemini 3.8 Flash measured |
| Neither model has a result on the other 13 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 40.9 | 43.7 | Step 5 Preview |
| Epoch Capabilities Index | 157.4 | Not measured | Only Gemini 3.8 Flash measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index92
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| AA Agentic Index | 40.2 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: agents building web apps | 1238 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: agents building full-stack apps | 1244 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: agents building mobile apps | 1236 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: agents building Android apps | 1078 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: agents building Godot games | 1273 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: PowerPoint slides | 1153 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: spreadsheets | 1022 | Not measured | Only Gemini 3.8 Flash measured |
| LiveBench research | 79.1% | 72.8% | Gemini 3.8 Flash |
| LiveBench browser | 64.0% | 24.0% | Gemini 3.8 Flash |
| LiveBench long agentic | 30.1% | 27.5% | Gemini 3.8 Flash |
| SREBench (Vals AI) | 11.1% | 0.4% | Gemini 3.8 Flash |
| Finance Agent (Vals AI) | 61.4% | 50.7% | Gemini 3.8 Flash |
| Tax Agent Bench (Vals AI) | 32.1% | 27.3% | Gemini 3.8 Flash |
| Legal Research Bench (Vals AI) | 38.9% | 40.4% | Step 5 Preview |
| Harvey's Legal Agent Benchmark (Vals AI) | 12.1% | Not measured | Only Gemini 3.8 Flash measured |
| Excel Modeling Benchmark (Vals AI) | 72.2% | 57.1% | Gemini 3.8 Flash |
| BioMysteryBench (Vals AI) | 62.2% | 67.4% | Step 5 Preview |
| Terminal-Bench-Science (Vals AI) | 8.6% | 2.9% | Gemini 3.8 Flash |
| Vals Science Index | 35.8% | 29.6% | Gemini 3.8 Flash |
| Time Horizon Index (Vals AI) | 18.8% | Not measured | Only Gemini 3.8 Flash measured |
| CUA-bench (Vals AI) | 4.2% | Not measured | Only Gemini 3.8 Flash measured |
| APEX-Agents | 64.3% | Not measured | Only Gemini 3.8 Flash measured |
| Remote Labor Index | 5.8% | Not measured | Only Gemini 3.8 Flash measured |
| Vending-Bench 2 | $5,094 | Not measured | Only Gemini 3.8 Flash measured |
| Neither model has a result on the other 16 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index80
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| GPQA Diamond | 94.4% | Not measured | Only Gemini 3.8 Flash measured |
| LiveBench reasoning | 89.3% | 83.8% | Gemini 3.8 Flash |
| MysteryMechanism (Vals AI) | 36.5% | 18.5% | Gemini 3.8 Flash |
| MMMU Pro (Vals AI) | 89.1% | Not measured | Only Gemini 3.8 Flash measured |
| GPQA Diamond (Artificial Analysis run) | 95.3% | Not measured | Only Gemini 3.8 Flash measured |
| Humanity's Last Exam (Artificial Analysis run) | 47.8% | 46.5% | Gemini 3.8 Flash |
| GPQA Diamond (Epoch AI run) | 95.4% | Not measured | Only Gemini 3.8 Flash measured |
| ARC-AGI | 98.5% | Not measured | Only Gemini 3.8 Flash measured |
| ARC-AGI-2 | 89.2% | Not measured | Only Gemini 3.8 Flash measured |
| Chess puzzles | 58.0% | Not measured | Only Gemini 3.8 Flash measured |
| Mystery game puzzles | 47.0% | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: hard prompts | 1509 | 1461 | Gemini 3.8 Flash |
| LMArena Text: expert | 1526 | 1485 | Gemini 3.8 Flash |
| LMArena Text: hard prompts in English | 1504 | 1454 | Gemini 3.8 Flash |
| LMArena Vision: diagrams | 1345 | 1308 | Gemini 3.8 Flash |
| LMArena Vision: homework | 1351 | 1265 | Gemini 3.8 Flash |
| Neither model has a result on the other 6 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| OTIS Mock AIME | 98.9% | Not measured | Only Gemini 3.8 Flash measured |
| FrontierMath, tiers 1 to 3 | 68.4% | Not measured | Only Gemini 3.8 Flash measured |
| FrontierMath, tier 4 | 22.0% | Not measured | Only Gemini 3.8 Flash measured |
| ProofBench | 48.0% | 42.0% | Gemini 3.8 Flash |
| LMArena Text: math | 1532 | 1458 | Gemini 3.8 Flash |
| LMArena Text: mathematical work | 1519 | 1460 | Gemini 3.8 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 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| MMLU Pro | 90.2% | Not measured | Only Gemini 3.8 Flash measured |
| LegalBench (Vals AI) | 87.0% | Not measured | Only Gemini 3.8 Flash measured |
| TaxEval (Vals AI) | 74.5% | Not measured | Only Gemini 3.8 Flash measured |
| MortgageTax (Vals AI) | 65.3% | Not measured | Only Gemini 3.8 Flash measured |
| MedCode (Vals AI) | 48.1% | Not measured | Only Gemini 3.8 Flash measured |
| SimpleQA Verified | 69.7% | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: science | 1523 | 1454 | Gemini 3.8 Flash |
| LMArena Text: medicine and healthcare | 1508 | 1439 | Gemini 3.8 Flash |
| LMArena Text: legal and government | 1497 | 1437 | Gemini 3.8 Flash |
| LMArena Text: business and finance | 1479 | 1435 | Gemini 3.8 Flash |
| FACTS Benchmark Suite | 69.4% | Not measured | Only Gemini 3.8 Flash measured |
| 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 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| Design Arena: websites | 1300 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: UI components | 1316 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: game dev | 1304 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: data viz | 1270 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: 3D | 1294 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: all code categories | 1303 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: SVG | 1296 | Not measured | Only Gemini 3.8 Flash measured |
| Design Arena: ASCII art | 1262 | Not measured | Only Gemini 3.8 Flash measured |
| LiveBench design | 72.7% | 76.9% | Step 5 Preview |
GeneralLike-for-likePublic boards, shown as published, kept out of the index182
| Benchmark | Gemini 3.8 Flash | Step 5 Preview | Leader |
|---|---|---|---|
| Vals Index | 56.0% | Not measured | Only Gemini 3.8 Flash measured |
| LiveBench data analysis | 54.0% | 63.5% | Step 5 Preview |
| Public Benefits Bench (Vals AI) | 65.3% | 58.1% | Gemini 3.8 Flash |
| SAGE (Vals AI) | 35.1% | Not measured | Only Gemini 3.8 Flash measured |
| MedScribe (Vals AI) | 84.5% | Not measured | Only Gemini 3.8 Flash measured |
| Long context reasoning (AA-LCR) | 84.0% | 88.3% | Step 5 Preview |
| DTBench | 95.7% | Not measured | Only Gemini 3.8 Flash measured |
| LMCA | 52.9% | Not measured | Only Gemini 3.8 Flash measured |
| Surface Evolver Bench | 76.9% | Not measured | Only Gemini 3.8 Flash measured |
| Furniture assembly | 31.7% | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text | 1499 | 1440 | Gemini 3.8 Flash |
| LMArena Text: instruction following | 1492 | 1438 | Gemini 3.8 Flash |
| LMArena Text: creative writing | 1488 | 1409 | Gemini 3.8 Flash |
| LMArena Vision | 1315 | 1265 | Gemini 3.8 Flash |
| LMArena Text: English | 1498 | 1447 | Gemini 3.8 Flash |
| LMArena Text: non-English | 1491 | 1428 | Gemini 3.8 Flash |
| LMArena Text: Chinese | 1553 | 1500 | Gemini 3.8 Flash |
| LMArena Text: German | 1501 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: Japanese | 1507 | 1383 | Gemini 3.8 Flash |
| LMArena Text: Korean | 1464 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: French | 1496 | 1438 | Gemini 3.8 Flash |
| LMArena Text: Spanish | 1487 | 1441 | Gemini 3.8 Flash |
| LMArena Text: Russian | 1514 | 1433 | Gemini 3.8 Flash |
| LMArena Text: Polish | 1497 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Text: multi-turn | 1503 | 1437 | Gemini 3.8 Flash |
| LMArena Text: longer queries | 1507 | 1449 | Gemini 3.8 Flash |
| LMArena Text: writing and language | 1494 | 1425 | Gemini 3.8 Flash |
| LMArena Text: entertainment and media | 1480 | 1412 | Gemini 3.8 Flash |
| LMArena Vision: English | 1318 | 1290 | Gemini 3.8 Flash |
| LMArena Vision: Chinese | 1416 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Vision: reading text | 1333 | 1282 | Gemini 3.8 Flash |
| LMArena Vision: humor | 1321 | Not measured | Only Gemini 3.8 Flash measured |
| LMArena Vision: creative writing | 1299 | Not measured | Only Gemini 3.8 Flash measured |
| Neither model has a result on the other 13 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemini 3.8 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 3.8 Flash costs less
Agent loop
200K input + 10K output tokens
Gemini 3.8 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 3.8 Flash | StepFun Step 5 Preview |
|---|---|---|
| Maker | StepFun | |
| Weights | No public weights listed | Open weights |
| Released | September 2, 2026 | October 8, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $0.75 | $1.00 |
| Output $/Mtok | $3.75 | $2.70 |
| Coding index | 61.6 | 61.1 |
| Independent measurements behind it | 6 | 4 |
| Cost per solved task | $2.74 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemini 3.8 Flash better than Step 5 Preview for coding?
Gemini 3.8 Flash and Step 5 Preview are level for coding, 61.6 against 61.1 on the Vibe Coding Index, so price and context decide. Pick Step 5 Preview for output tokens 1.4x cheaper ($2.70 vs $3.75 per million) and work measured by LiveBench coding or Terminal-Bench 4 (Vals AI), where it leads. Pick Gemini 3.8 Flash for work measured by LMArena WebDev or Vibe Code Bench (Vals AI), where it leads.
Which is cheaper, Gemini 3.8 Flash or Step 5 Preview?
Gemini 3.8 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 3.8 Flash only.
Which has the bigger context window, Gemini 3.8 Flash or Step 5 Preview?
Neither: both take 1.05M tokens.
How many benchmarks is the Gemini 3.8 Flash vs Step 5 Preview comparison based on?
61 benchmarks have a result for both models as of October 22, 2026. Gemini 3.8 Flash's coding index rests on 6 independent benchmarks and Step 5 Preview's on 4. Where only one model has a result, the page says so instead of naming a winner.