Gemini 4 Argon vs Step 5 Preview
Gemini 4 Argon leads Step 5 Preview on coding, 81.0 to 61.1 on the Vibe Coding Index, and Step 5 Preview costs 3.7x less per output token ($2.70 vs $10.00 per million). Updated October 22, 2026; 53 benchmarks have a result for both.
Gemini 4 Argon or Step 5 Preview: which to pick
Pick Gemini 4 Argon for coding: it scores 81.0 on the Vibe Coding Index against 61.1 for Step 5 Preview.
Pick Step 5 Preview for output tokens 3.7x cheaper ($2.70 vs $10.00 per million) and work measured by LiveBench design or Long context reasoning (AA-LCR), where it leads.
- Shared results
- 53
- Gemini 4 Argon only
- 41
- Step 5 Preview only
- 8
- 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
Gemini 4 Argon
Gemini 4 Argon scores 81.0 against 61.1.
5 and 4 benchmarks behind the scores
Intelligence
AA Intelligence Index
Gemini 4 Argon
Gemini 4 Argon leads, 52.6 against 43.7.
Same board, both measured
Web design
Design Arena: websites
No call
Only Gemini 4 Argon has a Design Arena: websites result, 1315.
Directional only
General
Vals Index
No call
Only Gemini 4 Argon has a Vals Index result, 69.2%.
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 4 Argon $0.0070, Step 5 Preview $0.0024 per turn.
List prices
At a glance
- Coding index
- Gemini 4 Argon 81.0 from 5 benchmarks, Step 5 Preview 61.1 from 4 benchmarks.
- Widest category
- Math, 28.8 points apart on average over 3 shared benchmarks.
- Widest single gap
- LiveBench browser: Gemini 4 Argon 92.0% against Step 5 Preview 24.0%, 68.0 points apart on a 0 to 100 scale.
- Output price
- Gemini 4 Argon $10.00 against Step 5 Preview $2.70 per million output tokens, 3.7x 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 4 Argon | Bars | Step 5 Preview | Ties |
|---|---|---|---|---|
| Coding | 12 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 11 | 0 | ||
| Reasoning | 6 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| Web design | 0 | 1 | ||
| General | 14 | 1 | ||
| All shared boards | 51 | 2 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index120
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench coding | 89.7% | 85.6% | Gemini 4 Argon |
| LMArena WebDev | 1678 | 1567 | Gemini 4 Argon |
| Terminal-Bench 4 (Vals AI) | 57.6% | 31.8% | Gemini 4 Argon |
| Vibe Code Bench (Vals AI) | 91.9% | 70.4% | Gemini 4 Argon |
| Code migration (Vals AI) | 68.2% | 39.0% | Gemini 4 Argon |
| ProgramBench (Vals AI) | 2.5% | Not measured | Only Gemini 4 Argon measured |
| IOI (Vals AI) | 100.0% | 38.1% | Gemini 4 Argon |
| Terminal-Bench 2.1 (Vals AI) | 57.7% | Not measured | Only Gemini 4 Argon measured |
| SWE-bench Multimodal (Vals AI) | 56.0% | 40.8% | Gemini 4 Argon |
| Vibe Code Bench 1-100 (Vals AI) | 21.2% | Not measured | Only Gemini 4 Argon measured |
| CyberBench (Vals AI) | 77.9% | Not measured | Only Gemini 4 Argon measured |
| SciCode | 61.8% | 58.9% | Gemini 4 Argon |
| FrontierCode | 47.5% | Not measured | Only Gemini 4 Argon measured |
| FrontierSWE | 55.0% | Not measured | Only Gemini 4 Argon measured |
| MirrorCode | 58.9% | Not measured | Only Gemini 4 Argon measured |
| ALE-Bench | 2100.1 | Not measured | Only Gemini 4 Argon measured |
| LMArena Text: coding | 1546 | 1485 | Gemini 4 Argon |
| LMArena Text: software and IT | 1556 | 1473 | Gemini 4 Argon |
| LMArena WebDev: React | 1692 | 1583 | Gemini 4 Argon |
| LMArena WebDev: HTML | 1643 | 1556 | Gemini 4 Argon |
| LMArena WebDev: from an image | Not measured | 1550 | Only Step 5 Preview measured |
| SciCode | 12.3% | Not measured | Only Gemini 4 Argon measured |
| Neither model has a result on the other 17 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 52.6 | 43.7 | Gemini 4 Argon |
| Epoch Capabilities Index | 166.4 | Not measured | Only Gemini 4 Argon measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index110
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| Design Arena: spreadsheets | 1157 | Not measured | Only Gemini 4 Argon measured |
| LiveBench research | 83.1% | 72.8% | Gemini 4 Argon |
| LiveBench browser | 92.0% | 24.0% | Gemini 4 Argon |
| LiveBench long agentic | 69.9% | 27.5% | Gemini 4 Argon |
| SREBench (Vals AI) | 44.3% | 0.4% | Gemini 4 Argon |
| Finance Agent (Vals AI) | 65.4% | 50.7% | Gemini 4 Argon |
| Tax Agent Bench (Vals AI) | 44.6% | 27.3% | Gemini 4 Argon |
| Legal Research Bench (Vals AI) | 54.8% | 40.4% | Gemini 4 Argon |
| Harvey's Legal Agent Benchmark (Vals AI) | 27.1% | Not measured | Only Gemini 4 Argon measured |
| Excel Modeling Benchmark (Vals AI) | 75.2% | 57.1% | Gemini 4 Argon |
| BioMysteryBench (Vals AI) | 76.3% | 67.4% | Gemini 4 Argon |
| Terminal-Bench-Science (Vals AI) | 44.3% | 2.9% | Gemini 4 Argon |
| Vals Science Index | 55.4% | 29.6% | Gemini 4 Argon |
| Time Horizon Index (Vals AI) | 52.5% | Not measured | Only Gemini 4 Argon measured |
| CUA-bench (Vals AI) | 4.8% | Not measured | Only Gemini 4 Argon measured |
| Robo-bench (Vals AI) | 0.8% | Not measured | Only Gemini 4 Argon measured |
| APEX-Agents | 82.4% | Not measured | Only Gemini 4 Argon measured |
| Vending-Bench 2 | $13,718 | Not measured | Only Gemini 4 Argon measured |
| Neither model has a result on the other 23 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench reasoning | 95.1% | 83.8% | Gemini 4 Argon |
| MysteryMechanism (Vals AI) | 45.5% | 18.5% | Gemini 4 Argon |
| Humanity's Last Exam (Artificial Analysis run) | 57.1% | 46.5% | Gemini 4 Argon |
| GPQA Diamond (Epoch AI run) | 96.4% | Not measured | Only Gemini 4 Argon measured |
| ARC-AGI | 98.5% | Not measured | Only Gemini 4 Argon measured |
| ARC-AGI-2 | 91.3% | Not measured | Only Gemini 4 Argon measured |
| Chess puzzles | 59.0% | Not measured | Only Gemini 4 Argon measured |
| Mystery game puzzles | 64.0% | Not measured | Only Gemini 4 Argon measured |
| EBR-bench | 64.8% | Not measured | Only Gemini 4 Argon measured |
| LMArena Text: hard prompts | 1546 | 1461 | Gemini 4 Argon |
| LMArena Text: expert | 1550 | 1485 | Gemini 4 Argon |
| LMArena Text: hard prompts in English | 1548 | 1454 | Gemini 4 Argon |
| LMArena Vision: diagrams | Not measured | 1308 | Only Step 5 Preview measured |
| LMArena Vision: homework | Not measured | 1265 | Only Step 5 Preview measured |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| OTIS Mock AIME | 100.0% | Not measured | Only Gemini 4 Argon measured |
| FrontierMath, tiers 1 to 3 | 86.0% | Not measured | Only Gemini 4 Argon measured |
| FrontierMath, tier 4 | 78.0% | Not measured | Only Gemini 4 Argon measured |
| ProofBench | 99.0% | 42.0% | Gemini 4 Argon |
| LMArena Text: math | 1536 | 1458 | Gemini 4 Argon |
| LMArena Text: mathematical work | 1557 | 1460 | Gemini 4 Argon |
| 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 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| LegalBench (Vals AI) | 88.3% | Not measured | Only Gemini 4 Argon measured |
| MedCode (Vals AI) | 58.8% | Not measured | Only Gemini 4 Argon measured |
| SimpleQA Verified | 76.0% | Not measured | Only Gemini 4 Argon measured |
| LMArena Text: science | 1548 | 1454 | Gemini 4 Argon |
| LMArena Text: medicine and healthcare | 1555 | 1439 | Gemini 4 Argon |
| LMArena Text: legal and government | 1548 | 1437 | Gemini 4 Argon |
| LMArena Text: business and finance | 1525 | 1435 | Gemini 4 Argon |
| Neither model has a result on the other 11 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| Design Arena: websites | 1315 | Not measured | Only Gemini 4 Argon measured |
| Design Arena: UI components | 1333 | Not measured | Only Gemini 4 Argon measured |
| Design Arena: game dev | 1377 | Not measured | Only Gemini 4 Argon measured |
| Design Arena: data viz | 1332 | Not measured | Only Gemini 4 Argon measured |
| Design Arena: 3D | 1388 | Not measured | Only Gemini 4 Argon measured |
| Design Arena: all code categories | 1329 | Not measured | Only Gemini 4 Argon measured |
| LiveBench design | 73.7% | 76.9% | Step 5 Preview |
| Neither model has a result on the other 2 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index141
| Benchmark | Gemini 4 Argon | Step 5 Preview | Leader |
|---|---|---|---|
| Vals Index | 69.2% | Not measured | Only Gemini 4 Argon measured |
| LiveBench data analysis | 85.2% | 63.5% | Gemini 4 Argon |
| Public Benefits Bench (Vals AI) | 69.8% | 58.1% | Gemini 4 Argon |
| SAGE (Vals AI) | 53.6% | Not measured | Only Gemini 4 Argon measured |
| MedScribe (Vals AI) | 87.4% | Not measured | Only Gemini 4 Argon measured |
| Long context reasoning (AA-LCR) | 79.7% | 88.3% | Step 5 Preview |
| DTBench | 97.6% | Not measured | Only Gemini 4 Argon measured |
| LMCA | 63.7% | Not measured | Only Gemini 4 Argon measured |
| Furniture assembly | 72.5% | Not measured | Only Gemini 4 Argon measured |
| LMArena Text | 1532 | 1440 | Gemini 4 Argon |
| LMArena Text: instruction following | 1534 | 1438 | Gemini 4 Argon |
| LMArena Text: creative writing | 1527 | 1409 | Gemini 4 Argon |
| LMArena Vision | Not measured | 1265 | Only Step 5 Preview measured |
| LMArena Text: English | 1538 | 1447 | Gemini 4 Argon |
| LMArena Text: non-English | 1522 | 1428 | Gemini 4 Argon |
| LMArena Text: Chinese | 1611 | 1500 | Gemini 4 Argon |
| LMArena Text: Japanese | Not measured | 1383 | Only Step 5 Preview measured |
| LMArena Text: French | Not measured | 1438 | Only Step 5 Preview measured |
| LMArena Text: Spanish | 1503 | 1441 | Gemini 4 Argon |
| LMArena Text: Russian | 1534 | 1433 | Gemini 4 Argon |
| LMArena Text: multi-turn | 1556 | 1437 | Gemini 4 Argon |
| LMArena Text: longer queries | 1547 | 1449 | Gemini 4 Argon |
| LMArena Text: writing and language | 1529 | 1425 | Gemini 4 Argon |
| LMArena Text: entertainment and media | 1527 | 1412 | Gemini 4 Argon |
| LMArena Vision: English | Not measured | 1290 | Only Step 5 Preview measured |
| LMArena Vision: reading text | Not measured | 1282 | Only Step 5 Preview measured |
| Neither model has a result on the other 20 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemini 4 Argon | 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
Step 5 Preview costs less
Agent loop
200K input + 10K output tokens
Step 5 Preview 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 4 Argon | StepFun Step 5 Preview |
|---|---|---|
| Maker | StepFun | |
| Weights | No public weights listed | Open weights |
| Released | October 13, 2026 | October 8, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $2.00 | $1.00 |
| Output $/Mtok | $10.00 | $2.70 |
| Coding index | 81.0 | 61.1 |
| Independent measurements behind it | 5 | 4 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemini 4 Argon better than Step 5 Preview for coding?
Pick Gemini 4 Argon for coding: it scores 81.0 on the Vibe Coding Index against 61.1 for Step 5 Preview. Pick Step 5 Preview for output tokens 3.7x cheaper ($2.70 vs $10.00 per million) and work measured by LiveBench design or Long context reasoning (AA-LCR), where it leads.
Which is cheaper, Gemini 4 Argon or Step 5 Preview?
Gemini 4 Argon costs $2.00 per million input tokens and $10.00 per million output tokens; Step 5 Preview costs $1.00 and $2.70. Cost per solved task is known for neither of them.
Which has the bigger context window, Gemini 4 Argon or Step 5 Preview?
Neither: both take 1.05M tokens.
How many benchmarks is the Gemini 4 Argon vs Step 5 Preview comparison based on?
53 benchmarks have a result for both models as of October 22, 2026. Gemini 4 Argon's coding index rests on 5 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.