Skip to main content
BenchLM
Advertise

Model comparison

DeepSeek V4 Flash vs Grok 4.3

Updated July 30, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

DeepSeek V4 Flash

DeepSeek

58.0/100

Estimated · Public rank #68

90% interval 46.5–69.5

Grok 4.3

xAI

64.2/100

Supported · Public rank #36

90% interval 54.5–74.0

Grok 4.3 has the higher public score estimate, 64.2 versus 57.98, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.

1 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.

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

    DeepSeek V4 Flash

    DeepSeek V4 Flash 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

    DeepSeek V4 Flash

    DeepSeek V4 Flash 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

    DeepSeek V4 Flash

    DeepSeek V4 Flash 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

    No shared weighted benchmark basis supports a winner.

    Confidence: limited

  • Agentic work

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

    Not enough matched evidence

    No shared weighted benchmark basis supports a winner.

    Confidence: limited

  • Long documents

    Prompts that approach the documented context limit

    No clear pick

    The documented context windows are equal.

    Confidence: documented

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
1
DeepSeek V4 Flash only
21
Grok 4.3 only
1
Like-for-like categories
0 / 8

Category results, on a stated basis

Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.

Agentic

Not comparable
DeepSeek V4 Flash
49.1
Grok 4.3
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Coding

Not comparable
DeepSeek V4 Flash
64.2
Grok 4.3
Not measured
Weighted basis
2 vs 0 rows
Reading
Not comparable

Reasoning

Not comparable
DeepSeek V4 Flash
Not measured
Grok 4.3
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Knowledge

Not comparable
DeepSeek V4 Flash
38.8
Grok 4.3
Not measured
Weighted basis
4 vs 0 rows
Reading
Not comparable

Math

Not comparable
DeepSeek V4 Flash
40.8
Grok 4.3
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Multilingual

Not comparable
DeepSeek V4 Flash
Not measured
Grok 4.3
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multimodal

Not comparable
DeepSeek V4 Flash
Not measured
Grok 4.3
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Instruction following

Not comparable
DeepSeek V4 Flash
Not measured
Grok 4.3
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

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.

A shared-evidence shape is not available.

BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.

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 Flash
$0.00028
Fits in one request
Grok 4.3
$0.0025
Fits in one request

DeepSeek V4 Flash has the lower modeled cost

Costs use the listed standard API rates.

Repository review

50K fresh input + 3K output tokens

DeepSeek V4 Flash
$0.00784
Fits in one request
Grok 4.3
$0.07
Fits in one request

DeepSeek V4 Flash 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 Flash
$0.00616
Fits in one request
Grok 4.3
$0.09
Fits in one request

DeepSeek V4 Flash 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.

DeepSeek V4 Flash

$0.0028 per 1M cached input tokens

Grok 4.3

$0.2 per 1M cached input tokens

Provider availability

DeepSeek V4 Flash

Generally Available · DeepSeek API, open weights

DeepSeek V4 release

Grok 4.3

Not sourced

Reasoning profile

DeepSeek V4 Flash

Non-Reasoning

Grok 4.3

Reasoning

Weight access

DeepSeek V4 Flash

Open Weight

Grok 4.3

Proprietary

License

DeepSeek V4 Flash

Open Weight

Grok 4.3

Proprietary

Release date

DeepSeek V4 Flash

2026-04-24

Grok 4.3

2026-04-30

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
Grok 4.3 has the higher public score estimate, 64.2 versus 57.98, but the 90% score intervals overlap.
Workload cost
Repository review: $0.00784 vs $0.07. Cache-heavy agent loop: $0.00616 vs $0.09.
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 evidence23 rows

Agentic

  • Terminal-Bench 2.0

    DeepSeek V4 Flash49.1%
    Source
    Grok 4.3—

    Not directly comparable

  • MCP Atlas

    DeepSeek V4 Flash64%
    Source
    Grok 4.3—

    Not directly comparable

  • Toolathlon

    DeepSeek V4 Flash40.7%
    Source
    Grok 4.3—

    Not directly comparable

  • Claw-Eval

    DeepSeek V4 Flash57.8%
    Source
    Grok 4.3—

    Not directly comparable

  • Gert Labs

    Shared source
    DeepSeek V4 Flash54.35%
    Grok 4.343.86%

    DeepSeek V4 Flash leads this result

  • ResearchClawBench

    DeepSeek V4 Flash—
    Grok 4.312.4%
    Source

    Not directly comparable

Coding

  • LiveCodeBench Pass@1-COT

    DeepSeek V4 Flash55.2%
    Source
    Grok 4.3—

    Not directly comparable

  • SWE-bench Verified

    DeepSeek V4 Flash73.7%
    Source
    Grok 4.3—

    Not directly comparable

  • SWE-bench Pro

    DeepSeek V4 Flash49.1%
    Source
    Grok 4.3—

    Not directly comparable

  • SWE Multilingual

    DeepSeek V4 Flash69.7%
    Source
    Grok 4.3—

    Not directly comparable

  • Terminal-Bench 2.0

    DeepSeek V4 Flash49.1%
    Source
    Grok 4.3—

    Not directly comparable

Reasoning

  • MRCR 1M

    DeepSeek V4 Flash37.5%
    Source
    Grok 4.3—

    Not directly comparable

  • CorpusQA 1M

    DeepSeek V4 Flash15.5%
    Source
    Grok 4.3—

    Not directly comparable

Knowledge

  • MMLU-Pro

    DeepSeek V4 Flash83%
    Source
    Grok 4.3—

    Not directly comparable

  • SimpleQA

    DeepSeek V4 Flash23.1%
    Source
    Grok 4.3—

    Not directly comparable

  • Chinese-SimpleQA

    DeepSeek V4 Flash71.5%
    Source
    Grok 4.3—

    Not directly comparable

  • GPQA

    DeepSeek V4 Flash71.2%
    Source
    Grok 4.3—

    Not directly comparable

  • GPQA-D

    DeepSeek V4 Flash71.2%
    Source
    Grok 4.3—

    Not directly comparable

  • HLE

    DeepSeek V4 Flash8.1%
    Source
    Grok 4.3—

    Not directly comparable

Math

  • HMMT Feb 2026

    DeepSeek V4 Flash40.8%
    Source
    Grok 4.3—

    Not directly comparable

  • IMOAnswerBench

    DeepSeek V4 Flash41.9%
    Source
    Grok 4.3—

    Not directly comparable

  • Apex

    DeepSeek V4 Flash1.0%
    Source
    Grok 4.3—

    Not directly comparable

  • Apex Shortlist

    DeepSeek V4 Flash9.3%
    Source
    Grok 4.3—

    Not directly comparable

Frequently asked questions

Which is better, DeepSeek V4 Flash or Grok 4.3?

Grok 4.3 has the higher public score estimate, 64.2 versus 57.98, 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 Flash or Grok 4.3?

The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.

Which is better for agentic tasks, DeepSeek V4 Flash or Grok 4.3?

The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.

Which costs less, DeepSeek V4 Flash or Grok 4.3?

For the stated presets, chat costs $0.00028 on DeepSeek V4 Flash and $0.0025 on Grok 4.3; repository review costs $0.00784 and $0.07; the cache-heavy agent loop costs $0.00616 and $0.09. Costs use the listed standard API rates.

Which has the larger context window, DeepSeek V4 Flash or Grok 4.3?

Both models list the same context window, 1M.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated July 30, 2026

Watch DeepSeek V4 Flash vs Grok 4.3

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

Read a sample issue

Join 2,000+ readers.