Skip to main content
BenchLM
Data

DeepSeek V4 Pro 0813 vs Step 5 Preview

Updated November 5, 2026. Rank says DeepSeek V4 Pro 0813 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Share or export
Share on XLinkedInAPI/MCP

Decision reading

DeepSeek V4 Pro 0813 has the higher public point estimate, 62.65 versus 56.81. Their conditional score ranges overlap. These ranges do not establish rank confidence. 11 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
DeepSeek logo
DeepSeek V4 Pro 0813

DeepSeek

62.65/100

Estimated · Public rank #42

Conditional range 48.3–77.0

Model B
StepFun logo
Step 5 Preview

StepFun

56.81/100

Estimated · Public rank #60

Conditional range 47.1–66.5

Shared results
11
DeepSeek V4 Pro 0813 only
31
Step 5 Preview only
13
Like-for-like categories
0 / 8
Estimated: DeepSeek V4 Pro 0813 and Step 5 Preview. Conditional ranges do not establish rank confidence.How 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

    Step 5 Preview is 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

    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.

47.6DeepSeek V4 Pro 081348.8Step 5 Preview

Directional only · BenchAlign v5.8

Step 5 Preview has the higher coding point estimate. Conditional score ranges do not establish rank confidence.

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.

3 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 v5.8 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.

Agentic

Directional only
DeepSeek V4 Pro 0813
53.7
Supported · #47/135
Step 5 Preview
55.3
Estimated · #42/135
Basis
BenchAlign v5.8 lane · 12 vs 13 public rows
Reading
Directional only

Coding

Directional only
DeepSeek V4 Pro 0813
47.6
Supported · #51/156
Step 5 Preview
48.8
Estimated · #47/156
Basis
BenchAlign v5.8 lane · 15 vs 6 public rows
Reading
Directional only

Knowledge

Directional only
DeepSeek V4 Pro 0813
63.1
Estimated · #41/177
Step 5 Preview
61.0
Supported · #44/177
Basis
BenchAlign v5.8 lane · 7 vs 2 public rows
Reading
Directional only

Reasoning

Not comparable
DeepSeek V4 Pro 0813
60.7
Unranked · 4 rankable rows
Step 5 Preview
82.5
Unranked · 2 rankable rows
Basis
Provisional lane · 1 vs 0 weighted rows
Reading
Not comparable

Multimodal

Not comparable
DeepSeek V4 Pro 0813
Not ranked
Step 5 Preview
73.3
#32/58
Basis
Provisional lane · 0 vs 2 weighted rows
Reading
Not comparable

Multilingual

Not comparable
DeepSeek V4 Pro 0813
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Instruction following

Not comparable
DeepSeek V4 Pro 0813
Not ranked
Step 5 Preview
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
DeepSeek V4 Pro 0813
87.4
Unranked · 4 rankable rows
Step 5 Preview
Not ranked
Basis
Provisional lane · 1 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 v5.8) 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

DeepSeek V4 Pro 0813
$0.0033
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

DeepSeek V4 Pro 0813
$0.07788
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

DeepSeek V4 Pro 0813
$0.0748
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.

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

DeepSeek V4 Pro 0813

Reasoning

Step 5 Preview

Reasoning

Weight access

DeepSeek V4 Pro 0813

Open Weight

Step 5 Preview

Open Weight

License

DeepSeek V4 Pro 0813

Open Weight

Step 5 Preview

Open Weight

Release date

DeepSeek V4 Pro 0813

2026-08-13

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
DeepSeek V4 Pro 0813 has the higher public point estimate, 62.65 versus 56.81. Their conditional score ranges overlap. These ranges do not establish rank confidence.
Workload cost
Repository review: $0.07788 vs $0.0581. Cache-heavy agent loop: $0.0748 vs $0.057.
Context tradeoff
Both models list 1M.
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, DeepSeek V4 Pro 0813 or Step 5 Preview?

DeepSeek V4 Pro 0813 has the higher public point estimate, 62.65 versus 56.81. Their conditional score ranges overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, DeepSeek V4 Pro 0813 or Step 5 Preview?

Step 5 Preview scores higher for coding on the public lane, 48.8 to 47.6. 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, DeepSeek V4 Pro 0813 or Step 5 Preview?

Step 5 Preview scores higher for agentic tasks on the public lane, 55.3 to 53.7. Step 5 Preview is 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, DeepSeek V4 Pro 0813 or Step 5 Preview?

For the stated presets, chat costs $0.0033 on DeepSeek V4 Pro 0813 and $0.00235 on Step 5 Preview; repository review costs $0.07788 and $0.0581; the cache-heavy agent loop costs $0.0748 and $0.057. Costs use the listed standard API rates.

Which has the larger context window, DeepSeek V4 Pro 0813 or Step 5 Preview?

Both models list the same context window, 1M.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

DeepSeek V4 Pro 0813
API / mo$3,960
Self-host / mo$37,843
Break-even597M/day
Step 5 Preview
API / mo$2,775
Self-host / mo$27,962
Break-even617M/day
Model the full break-even

Benchmark evidence

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

Browse raw public benchmark evidence55 rows

Agentic

  • Finch

    DeepSeek V4 Pro 081367.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.0

    DeepSeek V4 Pro 081367.9%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1

    DeepSeek V4 Pro 081387.9%
    Source
    Step 5 Preview85.0%
    Source

    DeepSeek V4 Pro 0813 leads this result

  • BrowseComp

    DeepSeek V4 Pro 081383.4%
    Source
    Step 5 Preview88.7%
    Source

    Step 5 Preview leads this result

  • HLE w/ tools

    DeepSeek V4 Pro 081360.0%
    Source
    Step 5 Preview—

    Not directly comparable

  • MCP Atlas

    DeepSeek V4 Pro 081373.6%
    Source
    Step 5 Preview85.6%
    Source

    Step 5 Preview leads this result

  • Toolathlon

    DeepSeek V4 Pro 081351.8%
    Source
    Step 5 Preview—

    Not directly comparable

  • CyberGym

    DeepSeek V4 Pro 081383.3%
    Source
    Step 5 Preview84.7%
    Source

    Step 5 Preview leads this result

  • Toolathlon-Verified

    DeepSeek V4 Pro 081374.1%
    Source
    Step 5 Preview74.1%
    Source

    Tie

  • Agents' Last Exam

    DeepSeek V4 Pro 081325.7%
    Source
    Step 5 Preview29.5%
    Source

    Step 5 Preview leads this result

  • AutomationBench

    DeepSeek V4 Pro 081331.8%
    Source
    Step 5 Preview44.0%
    Source

    Step 5 Preview leads this result

  • Terminal-Bench 2.1 (Vals)

    DeepSeek V4 Pro 081354.7%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 4.0

    DeepSeek V4 Pro 0813—
    Step 5 Preview33.30%
    Source

    Not directly comparable

  • JobBench

    DeepSeek V4 Pro 0813—
    Step 5 Preview59.0%
    Source

    Not directly comparable

  • APEX-Agents

    DeepSeek V4 Pro 0813—
    Step 5 Preview37.8%
    Source

    Not directly comparable

  • DRACO

    DeepSeek V4 Pro 0813—
    Step 5 Preview83.3%
    Source

    Not directly comparable

  • Hermes Index

    DeepSeek V4 Pro 0813—
    Step 5 Preview32.8%
    Source

    Not directly comparable

  • Hermes Bench

    DeepSeek V4 Pro 0813—
    Step 5 Preview69.2%
    Source

    Not directly comparable

Coding

  • LiveCodeBench Pass@1-COT

    DeepSeek V4 Pro 081393.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • Codeforces

    DeepSeek V4 Pro 08133206.0
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE-bench Verified

    DeepSeek V4 Pro 081380.6%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE-bench Pro

    DeepSeek V4 Pro 081355.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE Multilingual

    DeepSeek V4 Pro 081376.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.0

    DeepSeek V4 Pro 081367.9%
    Source
    Step 5 Preview—

    Not directly comparable

  • Vibe Code Bench

    DeepSeek V4 Pro 081349.93%
    Source
    Step 5 Preview—

    Not directly comparable

  • Terminal-Bench 2.1

    DeepSeek V4 Pro 081387.9%
    Source
    Step 5 Preview85.0%
    Source

    DeepSeek V4 Pro 0813 leads this result

  • NL2Repo

    DeepSeek V4 Pro 081361.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • DeepSWE

    DeepSeek V4 Pro 081362.7%
    Source
    Step 5 Preview67.7%
    Source

    Step 5 Preview leads this result

  • DSBench-FullStack

    DeepSeek V4 Pro 081371.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • DSBench-Hard

    DeepSeek V4 Pro 081367.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • OpenHarmony Bench

    DeepSeek V4 Pro 081359.0%
    Source
    Step 5 Preview—

    Not directly comparable

  • LiveCodeBench (Vals)

    DeepSeek V4 Pro 081387.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • SWE-bench (Vals)

    DeepSeek V4 Pro 081396.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • SciCode

    DeepSeek V4 Pro 0813—
    Step 5 Preview58.9%
    Source

    Not directly comparable

  • ProgramBench

    DeepSeek V4 Pro 0813—
    Step 5 Preview80.5%
    Source

    Not directly comparable

  • sweMarathon

    DeepSeek V4 Pro 0813—
    Step 5 Preview72.7%
    Source

    Not directly comparable

  • MLS-Bench Lite

    DeepSeek V4 Pro 0813—
    Step 5 Preview40.5%
    Source

    Not directly comparable

Reasoning

  • MRCR 1M

    DeepSeek V4 Pro 081383.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • CorpusQA 1M

    DeepSeek V4 Pro 081362.0%
    Source
    Step 5 Preview—

    Not directly comparable

  • ARC-AGI-1

    DeepSeek V4 Pro 081390.00%
    Source
    Step 5 Preview—

    Not directly comparable

  • ARC-AGI-2

    DeepSeek V4 Pro 081361.3%
    Source
    Step 5 Preview—

    Not directly comparable

  • CritPt

    DeepSeek V4 Pro 0813—
    Step 5 Preview20.9%
    Source

    Not directly comparable

Multimodal

  • OfficeQA Pro

    DeepSeek V4 Pro 0813—
    Step 5 Preview60.3%
    Source

    Not directly comparable

  • MMMU-Pro

    DeepSeek V4 Pro 0813—
    Step 5 Preview76%
    Source

    Not directly comparable

Knowledge

  • MMLU-Pro

    DeepSeek V4 Pro 081387.5%
    Source
    Step 5 Preview—

    Not directly comparable

  • Chinese-SimpleQA

    DeepSeek V4 Pro 081384.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • GPQA

    DeepSeek V4 Pro 081390.1%
    Source
    Step 5 Preview—

    Not directly comparable

  • GPQA-D

    DeepSeek V4 Pro 081390.1%
    Source
    Step 5 Preview93.5%
    Source

    Step 5 Preview leads this result

  • HLE

    DeepSeek V4 Pro 081342.7%
    Source
    Step 5 Preview46.5%
    Source

    Step 5 Preview leads this result

  • GPQA Diamond (Vals)

    DeepSeek V4 Pro 081392.4%
    Source
    Step 5 Preview—

    Not directly comparable

  • MMLU-Pro (Vals)

    DeepSeek V4 Pro 081387.0%
    Source
    Step 5 Preview—

    Not directly comparable

Math

  • HMMT Feb 2026

    DeepSeek V4 Pro 081395.2%
    Source
    Step 5 Preview—

    Not directly comparable

  • IMOAnswerBench

    DeepSeek V4 Pro 081389.8%
    Source
    Step 5 Preview—

    Not directly comparable

  • Apex

    DeepSeek V4 Pro 081338.3%
    Source
    Step 5 Preview—

    Not directly comparable

  • Apex Shortlist

    DeepSeek V4 Pro 081390.2%
    Source
    Step 5 Preview—

    Not directly comparable

55 public results · 11 shared

Related comparisons

Compare API pricingRead the methodologyOpen the model selector

Watch DeepSeek V4 Pro 0813 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 6,100+ subscribers.

Last updated November 5, 2026