Skip to main content
Radar

Every change to the models you run, with its source and its date. Releases, price changes, retirements, API changes, and incidents.Every change to the models you run, with its source.

Follow model changes
BenchLM

Claude Sonnet 5 vs Step 5 Preview

Decision reading

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.

2 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Anthropic logo
Model A
Claude Sonnet 5

Anthropic

69.56/100

Supported · Public rank #21

90% interval 66.1–73.0

StepFun logo
Model B
Step 5 Preview

StepFun

—

Evidence status unavailable

90% interval unavailable

Updated September 21, 2026. We do not rank this pair: at least one has no public score. Public scores include evidence status and uncertainty.

Share or export

Share on XLinkedInSocial cardCSVJSON

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 not ranked on the public lane for coding, so no winner is named for coding.

    Confidence: limited

  • Agentic work

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

    Not enough matched evidence

    Step 5 Preview is not ranked on the public lane for agentic, so no winner is named for agentic.

    Confidence: limited

  • Long documents

    Prompts that approach the documented context limit

    No clear pick

    The documented context windows are equal.

    Confidence: documented

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.

64.6Claude Sonnet 5—Step 5 Preview

Not comparable · BenchAlign

The coding row is not comparable on the public lane: at least one model is not measured or not ranked there.

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

Open the coding leaderboardShortlist models for this job

What is actually comparable

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

Shared results
2
Claude Sonnet 5 only
24
Step 5 Preview only
20
Like-for-like categories
0 / 8

1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign 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.

Multimodal

Directional only
Claude Sonnet 5
77.4
#14/49
Step 5 Preview
61.2
#28/49
Basis
Provisional lane · 1 vs 2 weighted rows
Reading
Directional only

Agentic

Not comparable
Claude Sonnet 5
66.2
Supported · #11/154
Step 5 Preview
Not ranked
Basis
BenchAlign lane · 7 vs 11 public rows
Reading
Not comparable

Coding

Not comparable
Claude Sonnet 5
64.6
Supported · #13/156
Step 5 Preview
Not ranked
Basis
BenchAlign lane · 11 vs 6 public rows
Reading
Not comparable

Reasoning

Not comparable
Claude Sonnet 5
77.4
Unranked · 2 rankable rows
Step 5 Preview
78.5
Unranked · 2 rankable rows
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Knowledge

Not comparable
Claude Sonnet 5
66.8
Supported · #20/186
Step 5 Preview
Not ranked
Basis
BenchAlign lane · 6 vs 2 public rows
Reading
Not comparable

Multilingual

Not comparable
Claude Sonnet 5
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Instruction following

Not comparable
Claude Sonnet 5
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
Claude Sonnet 5
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) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

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.

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

Claude Sonnet 5
$0.007
Fits in one request
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

Claude Sonnet 5
$0.13
Fits in one request
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

Claude Sonnet 5
$0.18
Fits in one request
Step 5 Preview
$0.057
Fits in one request

Step 5 Preview has the lower modeled cost

Costs use the listed standard API rates.

Specification differences

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

Cached-input rate

A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.

Claude Sonnet 5

$0.2 per 1M cached input tokens

Claude API pricing

Step 5 Preview

$0.05 per 1M cached input tokens

StepFun pricing and rate limits

Reasoning profile

Claude Sonnet 5

Reasoning

Step 5 Preview

Reasoning

Weight access

Claude Sonnet 5

Proprietary

Step 5 Preview

Pending

License

Claude Sonnet 5

Proprietary

Step 5 Preview

Pending

Release date

Claude Sonnet 5

2026-06-30

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
At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.
Workload cost
Repository review: $0.13 vs $0.0581. Cache-heavy agent loop: $0.18 vs $0.057.
Context tradeoff
Both models list 1M.

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

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

Browse raw public benchmark evidence46 rows

Agentic

  • Terminal-Bench 3.0

    Claude Sonnet 514.6%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Sonnet 580.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • BrowseComp

    Claude Sonnet 584.7%
    Source
    Step 5 Preview88.7%
    Source

    Step 5 Preview leads this result

  • HLE w/ tools

    Claude Sonnet 557.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • OSWorld-Verified

    Claude Sonnet 581.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    Claude Sonnet 574.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • ApprenticeBench

    Claude Sonnet 516%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Sonnet 5—
    Step 5 Preview85.0%
    Source

    Not directly comparable

  • Terminal-Bench 4.0

    Claude Sonnet 5—
    Step 5 Preview33.30%
    Source

    Not directly comparable

  • CyberGym

    Claude Sonnet 5—
    Step 5 Preview84.7%
    Source

    Not directly comparable

  • AutomationBench

    Claude Sonnet 5—
    Step 5 Preview44.0%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Claude Sonnet 5—
    Step 5 Preview74.1%
    Source

    Not directly comparable

  • MCP Atlas

    Claude Sonnet 5—
    Step 5 Preview85.6%
    Source

    Not directly comparable

  • JobBench

    Claude Sonnet 5—
    Step 5 Preview59.0%
    Source

    Not directly comparable

  • APEX-Agents

    Claude Sonnet 5—
    Step 5 Preview37.8%
    Source

    Not directly comparable

  • DRACO

    Claude Sonnet 5—
    Step 5 Preview83.3%
    Source

    Not directly comparable

  • Agents' Last Exam

    Claude Sonnet 5—
    Step 5 Preview29.5%
    Source

    Not directly comparable

Coding

  • SWE-bench Verified

    Claude Sonnet 585.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE-bench Pro

    Claude Sonnet 563.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE Multilingual

    Claude Sonnet 578.3%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE Multimodal

    Claude Sonnet 528.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Sonnet 580.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • FrontierCode 1.1 Main

    Claude Sonnet 542.7%
    Source
    Step 5 Preview—

    Not directly comparable

  • cursorBench32

    Claude Sonnet 561.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • VulcanBench CII v1

    Claude Sonnet 589.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • LiveCodeBench (Vals)

    Claude Sonnet 582.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE-bench (Vals)

    Claude Sonnet 579.6%
    Source
    Step 5 Preview—

    Not directly comparable

  • cursorBench40

    Claude Sonnet 534.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • DeepSWE

    Claude Sonnet 5—
    Step 5 Preview67.7%
    Source

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Sonnet 5—
    Step 5 Preview85.0%
    Source

    Not directly comparable

  • SciCode

    Claude Sonnet 5—
    Step 5 Preview58.9%
    Source

    Not directly comparable

  • ProgramBench

    Claude Sonnet 5—
    Step 5 Preview80.5%
    Source

    Not directly comparable

  • sweMarathon

    Claude Sonnet 5—
    Step 5 Preview72.7%
    Source

    Not directly comparable

  • MLS-Bench Lite

    Claude Sonnet 5—
    Step 5 Preview40.5%
    Source

    Not directly comparable

Reasoning

  • CritPt

    Claude Sonnet 5—
    Step 5 Preview20.9%
    Source

    Not directly comparable

Multimodal

  • CharXiv

    Claude Sonnet 588.3%
    Source
    Step 5 Preview—

    Not directly comparable

  • CharXiv w/o tools

    Claude Sonnet 577%
    Source
    Step 5 Preview—

    Not directly comparable

  • OfficeQA Pro

    Claude Sonnet 5—
    Step 5 Preview60.3%
    Source

    Not directly comparable

  • MMMU-Pro

    Claude Sonnet 5—
    Step 5 Preview76%
    Source

    Not directly comparable

Knowledge

  • HLE

    Claude Sonnet 557.4%
    Source
    Step 5 Preview46.5%
    Source

    Claude Sonnet 5 leads this result

  • HLE w/o tools

    Claude Sonnet 543.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • HLE-Verified

    Claude Sonnet 531.0%
    Source
    Step 5 Preview—

    Not directly comparable

  • LABBench2

    Claude Sonnet 580.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • GPQA Diamond (Vals)

    Claude Sonnet 588.9%
    Source
    Step 5 Preview—

    Not directly comparable

  • MMLU-Pro (Vals)

    Claude Sonnet 587.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • GPQA-D

    Claude Sonnet 5—
    Step 5 Preview93.5%
    Source

    Not directly comparable

Questions

Which is better, Claude Sonnet 5 or Step 5 Preview?

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Claude Sonnet 5 or Step 5 Preview?

Step 5 Preview is not ranked on the public lane for coding, so no winner is named for coding.

Which is better for agentic tasks, Claude Sonnet 5 or Step 5 Preview?

Step 5 Preview is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.

Which costs less, Claude Sonnet 5 or Step 5 Preview?

For the stated presets, chat costs $0.007 on Claude Sonnet 5 and $0.00235 on Step 5 Preview; repository review costs $0.13 and $0.0581; the cache-heavy agent loop costs $0.18 and $0.057. Costs use the listed standard API rates.

Which has the larger context window, Claude Sonnet 5 or Step 5 Preview?

Both models list the same context window, 1M.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated September 21, 2026

Watch Claude Sonnet 5 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.