Skip to main content
BenchLM

Gemini 4 Argon vs Step 5 Preview

Updated September 30, 2026. Rank says Step 5 Preview is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Share or export
Share on XLinkedInSocial cardCSVJSON

Decision reading

Step 5 Preview has the higher public score estimate, 68.55 versus 64.59, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 4 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
Google logo
Gemini 4 Argon

Google

64.59/100

Estimated · Public rank #32

90% interval 53.1–76.1

Model B
StepFun logo
Step 5 Preview

StepFun

68.55/100

Estimated · Public rank #19

90% interval 57.0–80.1

Shared results
4
Gemini 4 Argon only
13
Step 5 Preview only
18
Like-for-like categories
1 / 8
Estimated: Gemini 4 Argon and Step 5 PreviewHow the comparison works

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

  • Chat turn cost

    1K fresh input + 500 output tokens

    Step 5 Preview

    Step 5 Preview has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates

  • Cache-heavy agent loop cost

    200K cached + 20K fresh input + 10K output tokens

    Step 5 Preview

    Step 5 Preview has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates
Show secondary and unsupported calls

  • Repository review cost

    50K fresh input + 3K output tokens

    Step 5 Preview

    Step 5 Preview has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates

  • Coding work

    Code generation, repair, and software-engineering tasks

    Not enough matched evidence

    Step 5 Preview is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.

    Confidence: limited

  • Agentic work

    Tool use, computer use, and multi-step task completion

    Not enough matched evidence

    Gemini 4 Argon and Step 5 Preview are scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.

    Confidence: limited

  • Long documents

    Prompts that approach the documented context limit

    Not enough matched evidence

    A complete context comparison is not sourced.

    Confidence: limited

Which one for a specific job

Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.

The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.

68.4Gemini 4 Argon61.9Step 5 Preview

Directional only · BenchAlign v5.7

Gemini 4 Argon scores higher, but at least one score rests on Estimated evidence or a different benchmark set. Directional only, no winner.

Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.

Open the coding leaderboardShortlist models for this job
Same basis rules as the category table below

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

Bars run 0–100 on each benchmark’s normalized display scale

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign v5.7 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Knowledge

Like-for-like
Gemini 4 Argon
72.9
Supported · #11/170
Step 5 Preview
67.5
Supported · #19/170
Basis
BenchAlign v5.7 lane · 1 vs 2 public rows
Reading
Gemini 4 Argon leads · intervals overlap

Agentic

Directional only
Gemini 4 Argon
63.7
Estimated · #14/119
Step 5 Preview
63.3
Estimated · #15/119
Basis
BenchAlign v5.7 lane · 7 vs 11 public rows
Reading
Directional only

Coding

Directional only
Gemini 4 Argon
68.4
Supported · #8/144
Step 5 Preview
61.9
Estimated · #16/144
Basis
BenchAlign v5.7 lane · 4 vs 6 public rows
Reading
Directional only

Reasoning

Not comparable
Gemini 4 Argon
77.1
Unranked · 3 rankable rows
Step 5 Preview
79.6
Unranked · 2 rankable rows
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Multimodal

Not comparable
Gemini 4 Argon
Not ranked
Step 5 Preview
61.2
#28/49
Basis
Provisional lane · 0 vs 2 weighted rows
Reading
Not comparable

Multilingual

Not comparable
Gemini 4 Argon
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Instruction following

Not comparable
Gemini 4 Argon
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
Gemini 4 Argon
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.7) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

Supported evidence per lane · bars run 0–100Methodology

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

Gemini 4 Argon
$0.007
Fit state unavailable
Step 5 Preview
$0.00235
Fits in one request

Step 5 Preview has the lower modeled cost

Costs use the listed standard API rates.

Repository review

50K fresh input + 3K output tokens

Gemini 4 Argon
$0.13
Fit state unavailable
Step 5 Preview
$0.0581
Fits in one request

Step 5 Preview has the lower modeled cost

Costs use the listed standard API rates.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

Gemini 4 Argon
$0.16
Fit state unavailable
Step 5 Preview
$0.057
Fits in one request

Step 5 Preview has the lower modeled cost

Costs use the listed standard API rates.

Cached input falls back to the list input rate only where a cached rate is unpublished

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

Reasoning profile

Gemini 4 Argon

Reasoning

Step 5 Preview

Reasoning

Weight access

Gemini 4 Argon

Proprietary

Step 5 Preview

Pending

License

Gemini 4 Argon

Proprietary

Step 5 Preview

Pending

Release date

Gemini 4 Argon

2026-09-30

Step 5 Preview

2026-09-20

If you already use one of these models

Deployment change
The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal
Step 5 Preview has the higher public score estimate, 68.55 versus 64.59, but the 90% score intervals overlap.
Workload cost
Repository review: $0.13 vs $0.0581. Cache-heavy agent loop: $0.16 vs $0.057.
Context tradeoff
A complete documented context comparison is not available.
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, Gemini 4 Argon or Step 5 Preview?

Step 5 Preview has the higher public score estimate, 68.55 versus 64.59, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, Gemini 4 Argon or Step 5 Preview?

Gemini 4 Argon scores higher for coding on the public lane, 68.4 to 61.9. Step 5 Preview is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which is better for agentic tasks, Gemini 4 Argon or Step 5 Preview?

Gemini 4 Argon scores higher for agentic tasks on the public lane, 63.7 to 63.3. Gemini 4 Argon and Step 5 Preview are scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Gemini 4 Argon or Step 5 Preview?

For the stated presets, chat costs $0.007 on Gemini 4 Argon and $0.00235 on Step 5 Preview; repository review costs $0.13 and $0.0581; the cache-heavy agent loop costs $0.16 and $0.057. Costs use the listed standard API rates.

Which has the larger context window, Gemini 4 Argon or Step 5 Preview?

A complete documented context-window comparison is not available.

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence35 rows

Agentic

  • AutomationBench

    Gemini 4 Argon51.3%
    Source
    Step 5 Preview44.0%
    Source

    Gemini 4 Argon leads this result

  • Finance Agent v2

    Gemini 4 Argon65.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 4.0

    Gemini 4 Argon57.40%
    Source
    Step 5 Preview33.30%
    Source

    Gemini 4 Argon leads this result

  • Agents' Last Exam

    Gemini 4 Argon39.5%
    Source
    Step 5 Preview29.5%
    Source

    Gemini 4 Argon leads this result

  • OSWorld 2.0

    Gemini 4 Argon69.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • CWE-bench v1

    Gemini 4 Argon68.0%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench-Science 0.1 (6x verifier timeout)

    Gemini 4 Argon57.6%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1

    Gemini 4 Argon—
    Step 5 Preview85.0%
    Source

    Not directly comparable

  • CyberGym

    Gemini 4 Argon—
    Step 5 Preview84.7%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Gemini 4 Argon—
    Step 5 Preview74.1%
    Source

    Not directly comparable

  • MCP Atlas

    Gemini 4 Argon—
    Step 5 Preview85.6%
    Source

    Not directly comparable

  • JobBench

    Gemini 4 Argon—
    Step 5 Preview59.0%
    Source

    Not directly comparable

  • APEX-Agents

    Gemini 4 Argon—
    Step 5 Preview37.8%
    Source

    Not directly comparable

  • DRACO

    Gemini 4 Argon—
    Step 5 Preview83.3%
    Source

    Not directly comparable

  • BrowseComp

    Gemini 4 Argon—
    Step 5 Preview88.7%
    Source

    Not directly comparable

Coding

  • DeepSWE

    Gemini 4 Argon77.9%
    Source
    Step 5 Preview67.7%
    Source

    Gemini 4 Argon leads this result

  • FrontierSWE v2

    Gemini 4 Argon55.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • Vibe Code Bench

    Gemini 4 Argon91.90%
    Source
    Step 5 Preview—

    Not directly comparable

  • PostTrainBench v1.1

    Gemini 4 Argon45.3%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1

    Gemini 4 Argon—
    Step 5 Preview85.0%
    Source

    Not directly comparable

  • SciCode

    Gemini 4 Argon—
    Step 5 Preview58.9%
    Source

    Not directly comparable

  • ProgramBench

    Gemini 4 Argon—
    Step 5 Preview80.5%
    Source

    Not directly comparable

  • sweMarathon

    Gemini 4 Argon—
    Step 5 Preview72.7%
    Source

    Not directly comparable

  • MLS-Bench Lite

    Gemini 4 Argon—
    Step 5 Preview40.5%
    Source

    Not directly comparable

Reasoning

  • Graphwalks BFS 128K

    Gemini 4 Argon99.7%
    Source
    Step 5 Preview—

    Not directly comparable

  • GraphWalks BFS 256K–1M

    Gemini 4 Argon84.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • CritPt

    Gemini 4 Argon—
    Step 5 Preview20.9%
    Source

    Not directly comparable

Multimodal

  • Chartography (no tools)

    Gemini 4 Argon71.6%
    Source
    Step 5 Preview—

    Not directly comparable

  • LVBench

    Gemini 4 Argon91.7%
    Source
    Step 5 Preview—

    Not directly comparable

  • OfficeQA Pro

    Gemini 4 Argon—
    Step 5 Preview60.3%
    Source

    Not directly comparable

  • MMMU-Pro

    Gemini 4 Argon—
    Step 5 Preview76%
    Source

    Not directly comparable

Knowledge

  • LABBench2

    Gemini 4 Argon88.8%
    Source
    Step 5 Preview—

    Not directly comparable

  • GPQA-D

    Gemini 4 Argon—
    Step 5 Preview93.5%
    Source

    Not directly comparable

  • HLE

    Gemini 4 Argon—
    Step 5 Preview46.5%
    Source

    Not directly comparable

Instruction following

  • Gray Swan IPI (15 attempts)

    Gemini 4 Argon0.7%
    Source
    Step 5 Preview—

    Not directly comparable

35 public results · 4 shared

Related comparisons

Compare API pricingRead the methodologyOpen the model selector

Watch Gemini 4 Argon vs Step 5 Preview

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 2,000+ readers.

Last updated September 30, 2026