Skip to main content
BenchLM
Data

Claude Opus 4.6 (Adaptive) vs DeepSeek V4 Pro

Updated November 21, 2026. Rank says DeepSeek V4 Pro 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 has the higher public score estimate, 61.4 versus 59.8, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 7 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
Anthropic logo

Anthropic

59.8/100

Supported · Public rank #58

90% interval 55.8–63.8

Model B
DeepSeek logo

DeepSeek

61.4/100

Supported · Public rank #48

90% interval 57.0–65.7

API price per 1M tokens

Claude Opus 4.6 (Adaptive): No comparable API price

DeepSeek V4 Pro: $1.32 input / $3.96 output

Context window

Claude Opus 4.6 (Adaptive): 1M

DeepSeek V4 Pro: 1M

Released

Claude Opus 4.6 (Adaptive): February 1, 2026

DeepSeek V4 Pro: August 13, 2026

Shared results
7
Claude Opus 4.6 (Adaptive) only
0
DeepSeek V4 Pro only
38
Like-for-like categories
6 / 8
Supported: Claude Opus 4.6 (Adaptive) and DeepSeek V4 ProHow 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.

  • Coding work

    Code generation, repair, and software-engineering tasks

    Claude Opus 4.6 (Adaptive)

    Claude Opus 4.6 (Adaptive) has the higher public coding point estimate, 60.9 to 59.6, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

    Confidence: limited
  • Agentic work

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

    DeepSeek V4 Pro

    DeepSeek V4 Pro has the higher public agentic point estimate, 59.3 to 58.5, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

    Confidence: limited
Show secondary and unsupported calls
  • Long documents

    Prompts that approach the documented context limit

    No clear pick

    The documented context windows are equal.

    Confidence: documented
  • Chat turn cost

    1K fresh input + 500 output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates
  • Cache-heavy agent loop cost

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

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates
  • Repository review cost

    50K fresh input + 3K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

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.

60.9Claude Opus 4.6 (Adaptive)59.6DeepSeek V4 Pro

Like-for-like · BenchAlign v6.0

Claude Opus 4.6 (Adaptive) 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.

Same basis rules as the category table below

Or describe your own job

The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.

Describe the task and constraints. This selector recommends models; it cannot carry out the task or access your files and accounts.

We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy

What is actually comparable

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

1 category rests 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 v6.0 lane in every category, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence. Directional and not-comparable rows remain visible, but they never receive a winner.

Agentic

Like-for-like
Claude Opus 4.6 (Adaptive)
58.5
Supported · #63/179
DeepSeek V4 Pro
59.3
Supported · #60/179
Basis
BenchAlign v6.0 lane · 0 vs 13 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Coding

Like-for-like
Claude Opus 4.6 (Adaptive)
60.9
Supported · #51/179
DeepSeek V4 Pro
59.6
Supported · #56/179
Basis
BenchAlign v6.0 lane · 3 vs 15 public rows
Reading
Claude Opus 4.6 (Adaptive) leads · intervals overlap

Reasoning

Like-for-like
Claude Opus 4.6 (Adaptive)
55.3
Supported · #60/161
DeepSeek V4 Pro
64.6
Supported · #33/161
Basis
BenchAlign v6.0 lane · 2 vs 4 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Knowledge

Like-for-like
Claude Opus 4.6 (Adaptive)
59.2
Supported · #51/158
DeepSeek V4 Pro
59.8
Supported · #46/158
Basis
BenchAlign v6.0 lane · 2 vs 7 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Instruction following

Like-for-like
Claude Opus 4.6 (Adaptive)
56.9
Supported · #46/143
DeepSeek V4 Pro
56.5
Supported · #47/143
Basis
BenchAlign v6.0 lane · 0 vs 0 public rows
Reading
Practical tie

Math

Like-for-like
Claude Opus 4.6 (Adaptive)
57.8
Supported · #41/125
DeepSeek V4 Pro
59.4
Supported · #34/125
Basis
BenchAlign v6.0 lane · 0 vs 6 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Multimodal

Directional only
Claude Opus 4.6 (Adaptive)
56.4
Supported · #60/119
DeepSeek V4 Pro
60.6
Estimated · #55/119
Basis
BenchAlign v6.0 lane · 0 vs 0 public rows
Reading
Directional only

Multilingual

Not comparable
Claude Opus 4.6 (Adaptive)
Not ranked
DeepSeek V4 Pro
57.5
Estimated · #31/87
Basis
BenchAlign v6.0 lane · 0 vs 0 public rows
Reading
Not comparable

Ranks count the models scored in each category’s BenchAlign v6.0 lane, so denominators differ by category. 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

Claude Opus 4.6 (Adaptive)
API rate not published
Fits in one request
DeepSeek V4 Pro
$0.0033
Fits in one request

Claude Opus 4.6 (Adaptive) has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Claude Opus 4.6 (Adaptive)
API rate not published
Fits in one request
DeepSeek V4 Pro
$0.07788
Fits in one request

Claude Opus 4.6 (Adaptive) has no comparable published API token rate.

Cache-heavy agent loop

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

Claude Opus 4.6 (Adaptive)
API rate not published
Fits in one request
Cached-input rate unavailable
DeepSeek V4 Pro
$0.0748
Fits in one request

Claude Opus 4.6 (Adaptive) has no comparable published API token rate.

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

Price against score, among all priced models

Overall score against output price per 1M tokens, one model per family. Pricing as of November 20, 2026.

DeepSeek V4 Pro

20406080100$0.10$0.50$1$5$10$50$100Output price, $ per 1M tokens (log)Overall score
  • Claude Opus 4.6 (Adaptive) has no public API price, so it is not plotted.
  • 8 cheaper models score at least as high as DeepSeek V4 Pro. The nearest is Gemini 3.8 Flash (score 67.7, $3.75 per 1M output tokens).

Gray dots are the other priced models. The dark step line is the efficiency frontier: for a model under it, another model costs no more and scores at least as high.

See DeepSeek V4 Pro on the full price vs performance chart

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 Opus 4.6 (Adaptive)

No comparable hosted API rate

DeepSeek V4 Pro

$0.044 per 1M cached input tokens

DeepSeek: Models & Pricing

Reasoning profile

Claude Opus 4.6 (Adaptive)

Reasoning

DeepSeek V4 Pro

Reasoning

Weight access

Claude Opus 4.6 (Adaptive)

Proprietary

DeepSeek V4 Pro

Open Weight

License

Claude Opus 4.6 (Adaptive)

Proprietary

DeepSeek V4 Pro

Open Weight

Release date

Claude Opus 4.6 (Adaptive)

2026-02-01

DeepSeek V4 Pro

2026-08-13

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 has the higher public score estimate, 61.4 versus 59.8, but the 90% score intervals overlap.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
Both models list 1M.
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, Claude Opus 4.6 (Adaptive) or DeepSeek V4 Pro?

DeepSeek V4 Pro has the higher public score estimate, 61.4 versus 59.8, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, Claude Opus 4.6 (Adaptive) or DeepSeek V4 Pro?

Claude Opus 4.6 (Adaptive) has the higher public coding point estimate, 60.9 to 59.6, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which is better for agentic tasks, Claude Opus 4.6 (Adaptive) or DeepSeek V4 Pro?

DeepSeek V4 Pro has the higher public agentic tasks point estimate, 59.3 to 58.5, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which costs less, Claude Opus 4.6 (Adaptive) or DeepSeek V4 Pro?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, Claude Opus 4.6 (Adaptive) or DeepSeek V4 Pro?

Both models list the same context window, 1M.

Self-host vs API cost

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

Claude Opus 4.6 (Adaptive)
API / moNo API price on file
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
DeepSeek V4 Pro
API / mo$3,960
Self-host / mo$37,843
Break-evenNot reached at 70% load
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 evidence45 rows

Agentic

  • Terminal-Bench 4.1

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro6.70%
    Source

    Not directly comparable

  • Finch

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro67.4%
    Source

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro67.9%
    Source

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro87.9%
    Source

    Not directly comparable

  • BrowseComp

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro83.4%
    Source

    Not directly comparable

  • HLE w/ tools

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro60.0%
    Source

    Not directly comparable

  • MCP Atlas

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro73.6%
    Source

    Not directly comparable

  • Toolathlon

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro51.8%
    Source

    Not directly comparable

  • CyberGym

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro83.3%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro74.1%
    Source

    Not directly comparable

  • Agents' Last Exam

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro25.7%
    Source

    Not directly comparable

  • AutomationBench

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro31.8%
    Source

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro54.7%
    Source

    Not directly comparable

Coding

  • Vibe Code Bench

    Shared source
    Claude Opus 4.6 (Adaptive)53.50%
    DeepSeek V4 Pro49.93%

    Claude Opus 4.6 (Adaptive) leads this result

  • LiveCodeBench (Vals)

    Claude Opus 4.6 (Adaptive)84.7%
    Source
    DeepSeek V4 Pro87.5%
    Source

    DeepSeek V4 Pro leads this result

  • SWE-bench (Vals)

    Claude Opus 4.6 (Adaptive)78.2%
    Source
    DeepSeek V4 Pro96.4%
    Source

    DeepSeek V4 Pro leads this result

  • LiveCodeBench Pass@1-COT

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro93.5%
    Source

    Not directly comparable

  • Codeforces

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro3206.0
    Source

    Not directly comparable

  • SWE-bench Verified

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro80.6%
    Source

    Not directly comparable

  • SWE-bench Pro

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro55.4%
    Source

    Not directly comparable

  • SWE Multilingual

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro76.2%
    Source

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro67.9%
    Source

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro87.9%
    Source

    Not directly comparable

  • NL2Repo

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro61.5%
    Source

    Not directly comparable

  • DeepSWE

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro62.7%
    Source

    Not directly comparable

  • DSBench-FullStack

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro71.1%
    Source

    Not directly comparable

  • DSBench-Hard

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro67.2%
    Source

    Not directly comparable

  • OpenHarmony Bench v1.0

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro59.0%
    Source

    Not directly comparable

Reasoning

  • ARC-AGI-1

    Claude Opus 4.6 (Adaptive)93.00%
    Source
    DeepSeek V4 Pro90.00%
    Source

    Claude Opus 4.6 (Adaptive) leads this result

  • ARC-AGI-2

    Claude Opus 4.6 (Adaptive)68.8%
    Source
    DeepSeek V4 Pro61.3%
    Source

    Claude Opus 4.6 (Adaptive) leads this result

  • MRCR 1M

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro83.5%
    Source

    Not directly comparable

  • CorpusQA 1M

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro62.0%
    Source

    Not directly comparable

Knowledge

  • GPQA Diamond (Vals)

    Claude Opus 4.6 (Adaptive)89.6%
    Source
    DeepSeek V4 Pro92.4%
    Source

    DeepSeek V4 Pro leads this result

  • MMLU-Pro (Vals)

    Claude Opus 4.6 (Adaptive)89.1%
    Source
    DeepSeek V4 Pro87.0%
    Source

    Claude Opus 4.6 (Adaptive) leads this result

  • MMLU-Pro

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro87.5%
    Source

    Not directly comparable

  • Chinese-SimpleQA

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro84.4%
    Source

    Not directly comparable

  • GPQA

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro90.1%
    Source

    Not directly comparable

  • GPQA-D

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro90.1%
    Source

    Not directly comparable

  • HLE

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro42.7%
    Source

    Not directly comparable

Math

  • HMMT Feb 2026

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro95.2%
    Source

    Not directly comparable

  • IMOAnswerBench

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro89.8%
    Source

    Not directly comparable

  • Apex

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro38.3%
    Source

    Not directly comparable

  • Apex Shortlist

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro90.2%
    Source

    Not directly comparable

  • FrontierMath v2 (Tiers 1-3)

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro64.561%
    Source

    Not directly comparable

  • FrontierMath v2 (Tier 4)

    Claude Opus 4.6 (Adaptive)—
    DeepSeek V4 Pro26.829%
    Source

    Not directly comparable

45 public results · 7 shared

Watch Claude Opus 4.6 (Adaptive) vs DeepSeek V4 Pro

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

Read a sample issue

Join 7,000+ subscribers.

Last updated November 21, 2026