Skip to main content
BenchLM
Data

DeepSeek V4 Pro vs Gemini 3.1 Pro

Updated November 20, 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, 61.5 versus 61, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 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

61.5/100

Supported · Public rank #48

90% interval 57.1–65.8

Model B
Google logo

Google

61/100

Supported · Public rank #50

90% interval 57.0–65.0

API price per 1M tokens

DeepSeek V4 Pro: $1.32 input / $3.96 output

Gemini 3.1 Pro: $2 input / $12 output

Context window

DeepSeek V4 Pro: 1M

Gemini 3.1 Pro: 1M

Released

DeepSeek V4 Pro: August 13, 2026

Gemini 3.1 Pro: February 19, 2026

Shared results
11
DeepSeek V4 Pro only
34
Gemini 3.1 Pro only
24
Like-for-like categories
6 / 8
Supported: DeepSeek V4 Pro and Gemini 3.1 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

    DeepSeek V4 Pro

    DeepSeek V4 Pro has the higher public coding point estimate, 59.5 to 55.8, 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 57.6, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

    Confidence: limited
  • Chat turn cost

    1K fresh input + 500 output tokens

    DeepSeek V4 Pro

    DeepSeek V4 Pro 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
  • Cache-heavy agent loop cost

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

    DeepSeek V4 Pro

    DeepSeek V4 Pro has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates
  • Repository review cost

    50K fresh input + 3K output tokens

    DeepSeek V4 Pro

    DeepSeek V4 Pro has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates
  • 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.

59.5DeepSeek V4 Pro55.8Gemini 3.1 Pro

Like-for-like · BenchAlign v6.0

DeepSeek V4 Pro 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
DeepSeek V4 Pro
59.3
Supported · #60/178
Gemini 3.1 Pro
57.6
Supported · #69/178
Basis
BenchAlign v6.0 lane · 13 vs 8 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Coding

Like-for-like
DeepSeek V4 Pro
59.5
Supported · #55/178
Gemini 3.1 Pro
55.8
Supported · #63/178
Basis
BenchAlign v6.0 lane · 15 vs 9 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Reasoning

Like-for-like
DeepSeek V4 Pro
64.6
Supported · #33/161
Gemini 3.1 Pro
62.8
Supported · #37/161
Basis
BenchAlign v6.0 lane · 4 vs 2 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Knowledge

Like-for-like
DeepSeek V4 Pro
60.2
Supported · #44/157
Gemini 3.1 Pro
68.9
Supported · #20/157
Basis
BenchAlign v6.0 lane · 7 vs 6 public rows
Reading
Gemini 3.1 Pro leads · intervals overlap

Instruction following

Like-for-like
DeepSeek V4 Pro
56.5
Supported · #46/142
Gemini 3.1 Pro
55.9
Supported · #48/142
Basis
BenchAlign v6.0 lane · 0 vs 0 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Math

Like-for-like
DeepSeek V4 Pro
59.5
Supported · #33/124
Gemini 3.1 Pro
58.3
Supported · #35/124
Basis
BenchAlign v6.0 lane · 6 vs 2 public rows
Reading
DeepSeek V4 Pro leads · intervals overlap

Multimodal

Directional only
DeepSeek V4 Pro
60.7
Estimated · #52/116
Gemini 3.1 Pro
62.3
Supported · #42/116
Basis
BenchAlign v6.0 lane · 0 vs 8 public rows
Reading
Directional only

Multilingual

Not comparable
DeepSeek V4 Pro
57.5
Estimated · #31/87
Gemini 3.1 Pro
Not ranked
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

DeepSeek V4 Pro
$0.0033
Fits in one request
Gemini 3.1 Pro
$0.008
Fits in one request

DeepSeek V4 Pro has the lower modeled cost

Costs use the listed standard API rates.

Repository review

50K fresh input + 3K output tokens

DeepSeek V4 Pro
$0.07788
Fits in one request
Gemini 3.1 Pro
$0.136
Fits in one request

DeepSeek V4 Pro 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
$0.0748
Fits in one request
Gemini 3.1 Pro
$0.2
Fits in one request

DeepSeek V4 Pro has the lower modeled cost

Costs use the listed standard API rates.

Your workload

Enter a month of tokens to price it at both models’ listed API rates.

DeepSeek V4 Pro
$21.12
per month
Gemini 3.1 Pro
$44.00
per month

Standard listed rates, no batch or volume discounts.

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 ProGemini 3.1 Pro

20406080100$0.10$0.50$1$5$10$50$100Output price, $ per 1M tokens (log)Overall score
  • 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).
  • 15 cheaper models score at least as high as Gemini 3.1 Pro. The nearest is Gemini 4 Carbon (score 88.0, $10.00 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.

DeepSeek V4 Pro

$0.044 per 1M cached input tokens

DeepSeek: Models & Pricing

Gemini 3.1 Pro

$0.2 per 1M cached input tokens

Google Gemini API pricing

Reasoning profile

DeepSeek V4 Pro

Reasoning

Gemini 3.1 Pro

Reasoning

Weight access

DeepSeek V4 Pro

Open Weight

Gemini 3.1 Pro

Proprietary

License

DeepSeek V4 Pro

Open Weight

Gemini 3.1 Pro

Proprietary

Release date

DeepSeek V4 Pro

2026-08-13

Gemini 3.1 Pro

2026-02-19

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, 61.5 versus 61, but the 90% score intervals overlap.
Workload cost
Repository review: $0.07788 vs $0.136. Cache-heavy agent loop: $0.0748 vs $0.2.
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 or Gemini 3.1 Pro?

DeepSeek V4 Pro has the higher public score estimate, 61.5 versus 61, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, DeepSeek V4 Pro or Gemini 3.1 Pro?

DeepSeek V4 Pro has the higher public coding point estimate, 59.5 to 55.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which is better for agentic tasks, DeepSeek V4 Pro or Gemini 3.1 Pro?

DeepSeek V4 Pro has the higher public agentic tasks point estimate, 59.3 to 57.6, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which costs less, DeepSeek V4 Pro or Gemini 3.1 Pro?

For the stated presets, chat costs $0.0033 on DeepSeek V4 Pro and $0.008 on Gemini 3.1 Pro; repository review costs $0.07788 and $0.136; the cache-heavy agent loop costs $0.0748 and $0.2. Costs use the listed standard API rates.

Which has the larger context window, DeepSeek V4 Pro or Gemini 3.1 Pro?

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
API / mo$3,960
Self-host / mo$37,843
Break-evenNot reached at 70% load
Gemini 3.1 Pro
API / mo$10,500
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
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 evidence69 rows

Agentic

  • Terminal-Bench 4.1

    Shared source
    DeepSeek V4 Pro6.70%
    Gemini 3.1 Pro2.70%

    DeepSeek V4 Pro leads this result

  • Finch

    DeepSeek V4 Pro67.4%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Terminal-Bench 2.0

    DeepSeek V4 Pro67.9%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Terminal-Bench 2.1

    DeepSeek V4 Pro87.9%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • BrowseComp

    DeepSeek V4 Pro83.4%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • HLE w/ tools

    DeepSeek V4 Pro60.0%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • MCP Atlas

    DeepSeek V4 Pro73.6%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Toolathlon

    DeepSeek V4 Pro51.8%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • CyberGym

    DeepSeek V4 Pro83.3%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Toolathlon-Verified

    DeepSeek V4 Pro74.1%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Agents' Last Exam

    DeepSeek V4 Pro25.7%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • AutomationBench

    DeepSeek V4 Pro31.8%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    DeepSeek V4 Pro54.7%
    Source
    Gemini 3.1 Pro70.8%
    Source

    Gemini 3.1 Pro leads this result

  • Claw-Eval

    DeepSeek V4 Pro—
    Gemini 3.1 Pro57.8%
    Source

    Not directly comparable

  • DeepSearchQA

    DeepSeek V4 Pro—
    Gemini 3.1 Pro69.7%
    Source

    Not directly comparable

  • τ²-bench results

    DeepSeek V4 Pro—
    Gemini 3.1 Pro95.6%
    Source

    Not directly comparable

  • Gert Labs

    DeepSeek V4 Pro—
    Gemini 3.1 Pro56.87%
    Source

    Not directly comparable

  • ResearchClawBench

    DeepSeek V4 Pro—
    Gemini 3.1 Pro13.3%
    Source

    Not directly comparable

  • MazeBench

    DeepSeek V4 Pro—
    Gemini 3.1 Pro1%
    Source

    Not directly comparable

Coding

  • LiveCodeBench Pass@1-COT

    DeepSeek V4 Pro93.5%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Codeforces

    DeepSeek V4 Pro3206.0
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • SWE-bench Verified

    DeepSeek V4 Pro80.6%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • SWE-bench Pro

    DeepSeek V4 Pro55.4%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • SWE Multilingual

    DeepSeek V4 Pro76.2%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Terminal-Bench 2.0

    DeepSeek V4 Pro67.9%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Vibe Code Bench

    Shared source
    DeepSeek V4 Pro49.93%
    Gemini 3.1 Pro32.03%

    DeepSeek V4 Pro leads this result

  • Terminal-Bench 2.1

    DeepSeek V4 Pro87.9%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • NL2Repo

    DeepSeek V4 Pro61.5%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • DeepSWE

    DeepSeek V4 Pro62.7%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • DSBench-FullStack

    DeepSeek V4 Pro71.1%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • DSBench-Hard

    DeepSeek V4 Pro67.2%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • OpenHarmony Bench v1.0

    DeepSeek V4 Pro59.0%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • LiveCodeBench (Vals)

    DeepSeek V4 Pro87.5%
    Source
    Gemini 3.1 Pro88.5%
    Source

    Gemini 3.1 Pro leads this result

  • SWE-bench (Vals)

    DeepSeek V4 Pro96.4%
    Source
    Gemini 3.1 Pro78.8%
    Source

    DeepSeek V4 Pro leads this result

  • ScarfBench

    DeepSeek V4 Pro—
    Gemini 3.1 Pro37.6%
    Source

    Not directly comparable

  • LiveCodeBench Pro

    DeepSeek V4 Pro—
    Gemini 3.1 Pro82.9%
    Source

    Not directly comparable

  • React Native Evals

    DeepSeek V4 Pro—
    Gemini 3.1 Pro78.9%
    Source

    Not directly comparable

  • PostTrainBench v1.1

    DeepSeek V4 Pro—
    Gemini 3.1 Pro22.0%
    Source

    Not directly comparable

  • PostTrainBench

    DeepSeek V4 Pro—
    Gemini 3.1 Pro18.1%
    Source

    Not directly comparable

  • PostTrainBench v1

    DeepSeek V4 Pro—
    Gemini 3.1 Pro21.6%
    Source

    Not directly comparable

Reasoning

  • MRCR 1M

    DeepSeek V4 Pro83.5%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • CorpusQA 1M

    DeepSeek V4 Pro62.0%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • ARC-AGI-1

    DeepSeek V4 Pro90.00%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • ARC-AGI-2

    DeepSeek V4 Pro61.3%
    Source
    Gemini 3.1 Pro77.1%
    Source

    Gemini 3.1 Pro leads this result

  • ARC-AGI-3

    DeepSeek V4 Pro—
    Gemini 3.1 Pro0.4%
    Source

    Not directly comparable

Multimodal

  • ParseBench-1

    DeepSeek V4 Pro—
    Gemini 3.1 Pro67.32%
    Source

    Not directly comparable

  • MMMU-Pro

    DeepSeek V4 Pro—
    Gemini 3.1 Pro83.9%
    Source

    Not directly comparable

  • CharXiv

    DeepSeek V4 Pro—
    Gemini 3.1 Pro80.2%
    Source

    Not directly comparable

  • ERQA

    DeepSeek V4 Pro—
    Gemini 3.1 Pro69.4%
    Source

    Not directly comparable

  • SimpleVQA

    DeepSeek V4 Pro—
    Gemini 3.1 Pro72.4%
    Source

    Not directly comparable

  • ScreenSpot Pro

    DeepSeek V4 Pro—
    Gemini 3.1 Pro84.4%
    Source

    Not directly comparable

  • ZeroBench

    DeepSeek V4 Pro—
    Gemini 3.1 Pro29.0%
    Source

    Not directly comparable

  • MedXpertQA (MM)

    DeepSeek V4 Pro—
    Gemini 3.1 Pro81.3%
    Source

    Not directly comparable

Knowledge

  • MMLU-Pro

    DeepSeek V4 Pro87.5%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Chinese-SimpleQA

    DeepSeek V4 Pro84.4%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • GPQA

    DeepSeek V4 Pro90.1%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • GPQA-D

    DeepSeek V4 Pro90.1%
    Source
    Gemini 3.1 Pro94.3%
    Source

    Gemini 3.1 Pro leads this result

  • HLE

    DeepSeek V4 Pro42.7%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • GPQA Diamond (Vals)

    DeepSeek V4 Pro92.4%
    Source
    Gemini 3.1 Pro95.5%
    Source

    Gemini 3.1 Pro leads this result

  • MMLU-Pro (Vals)

    DeepSeek V4 Pro87.0%
    Source
    Gemini 3.1 Pro91.0%
    Source

    Gemini 3.1 Pro leads this result

  • HLE w/o tools

    DeepSeek V4 Pro—
    Gemini 3.1 Pro45.4%
    Source

    Not directly comparable

  • HealthBench Hard

    DeepSeek V4 Pro—
    Gemini 3.1 Pro20.6%
    Source

    Not directly comparable

  • MedXpertQA (Text)

    DeepSeek V4 Pro—
    Gemini 3.1 Pro71.5%
    Source

    Not directly comparable

Math

  • HMMT Feb 2026

    DeepSeek V4 Pro95.2%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • IMOAnswerBench

    DeepSeek V4 Pro89.8%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Apex

    DeepSeek V4 Pro38.3%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • Apex Shortlist

    DeepSeek V4 Pro90.2%
    Source
    Gemini 3.1 Pro—

    Not directly comparable

  • FrontierMath v2 (Tiers 1-3)

    Shared source
    DeepSeek V4 Pro64.561%
    Gemini 3.1 Pro59.649%

    DeepSeek V4 Pro leads this result

  • FrontierMath v2 (Tier 4)

    Shared source
    DeepSeek V4 Pro26.829%
    Gemini 3.1 Pro26.829%

    Tie

69 public results · 11 shared

Watch DeepSeek V4 Pro vs Gemini 3.1 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 20, 2026