Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM
Model A
Claude Sonnet 5

Anthropic

64.8/100

Estimated · Public rank #37

90% interval 50.5–79.1

Claude Sonnet 5 vs GPT-5.4 Pro

Updated August 15, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B
GPT-5.4 Pro

OpenAI

61.3/100

Estimated · Public rank #52

90% interval 44.4–78.2

Decision reading

Claude Sonnet 5 has the higher public score estimate, 64.78 versus 61.26, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.

3 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.

  • Long documents

    Prompts that approach the documented context limit

    GPT-5.4 Pro

    GPT-5.4 Pro has the larger documented context window.

    Confidence: documented

  • Chat turn cost

    1K fresh input + 500 output tokens

    Claude Sonnet 5

    Claude Sonnet 5 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

    Claude Sonnet 5

    Claude Sonnet 5 has the lower estimated token cost for this stated workload. GPT-5.4 Pro has no published cached-input rate, so cached tokens use its listed input rate.

    Confidence: rate-fallback

Show secondary and unsupported calls

  • Repository review cost

    50K fresh input + 3K output tokens

    Claude Sonnet 5

    Claude Sonnet 5 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

    The category averages use different weighted benchmark sets, so they are directional rather than like-for-like.

    Confidence: limited

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
3
Claude Sonnet 5 only
16
GPT-5.4 Pro only
8
Like-for-like categories
1 / 8

1 category uses different evidence sets. Those rows remain visible for coverage context but do not name a winner.

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.

Knowledge

Like-for-like
Claude Sonnet 5
57.4
GPT-5.4 Pro
58.7
Weighted basis
1 vs 1 rows
Reading
GPT-5.4 Pro leads

Agentic

Directional only
Claude Sonnet 5
81.9
GPT-5.4 Pro
89.3
Weighted basis
3 vs 1 rows
Reading
Directional only

Coding

Not comparable
Claude Sonnet 5
76.7
GPT-5.4 Pro
Not measured
Weighted basis
2 vs 0 rows
Reading
Not comparable

Reasoning

Not comparable
Claude Sonnet 5
Not measured
GPT-5.4 Pro
83.3
Weighted basis
0 vs 1 rows
Reading
Not comparable

Math

Not comparable
Claude Sonnet 5
Not measured
GPT-5.4 Pro
46.9
Weighted basis
0 vs 2 rows
Reading
Not comparable

Multilingual

Not comparable
Claude Sonnet 5
Not measured
GPT-5.4 Pro
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multimodal

Not comparable
Claude Sonnet 5
88.3
GPT-5.4 Pro
94.0
Weighted basis
1 vs 1 rows
Reading
Not comparable

Instruction following

Not comparable
Claude Sonnet 5
Not measured
GPT-5.4 Pro
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.

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
GPT-5.4 Pro
$0.12
Fits in one request

Claude Sonnet 5 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
GPT-5.4 Pro
$2.04
Fits in one request

Claude Sonnet 5 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
GPT-5.4 Pro
$8.40
Fits in one request
Cached input priced at the published list-input rate

Claude Sonnet 5 has the lower modeled cost

GPT-5.4 Pro has no published cached-input rate, so cached tokens use its listed input rate.

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

GPT-5.4 Pro

Not published

OpenAI pricing

Reasoning profile

Claude Sonnet 5

Reasoning

GPT-5.4 Pro

Reasoning

Weight access

Claude Sonnet 5

Proprietary

GPT-5.4 Pro

Proprietary

License

Claude Sonnet 5

Proprietary

GPT-5.4 Pro

Proprietary

Release date

Claude Sonnet 5

2026-06-30

GPT-5.4 Pro

2026-03-05

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
Claude Sonnet 5 has the higher public score estimate, 64.78 versus 61.26, but the 90% score intervals overlap.
Workload cost
Repository review: $0.13 vs $2.04. Cache-heavy agent loop: $0.18 vs $8.40.
Context tradeoff
GPT-5.4 Pro has the larger documented window (1.05M).

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 evidence27 rows

Agentic

  • Terminal-Bench 3.0

    Claude Sonnet 514.6%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Sonnet 580.4%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • BrowseComp

    Claude Sonnet 584.7%
    Source
    GPT-5.4 Pro89.3%
    Source

    GPT-5.4 Pro leads this result

  • HLE w/ tools

    Claude Sonnet 557.4%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • OSWorld-Verified

    Claude Sonnet 581.2%
    Source
    GPT-5.4 Pro—

    Not directly comparable

Coding

  • SWE-bench Verified

    Claude Sonnet 585.2%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • SWE-bench Pro

    Claude Sonnet 563.2%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • SWE Multilingual

    Claude Sonnet 578.3%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • SWE Multimodal

    Claude Sonnet 528.1%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • Terminal-Bench 2.0

    Claude Sonnet 580.4%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • FrontierCode 1.1 Main

    Claude Sonnet 542.7%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • cursorBench32

    Claude Sonnet 561.5%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • APEX-SWE

    Claude Sonnet 546.4%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • EEBench

    Claude Sonnet 540.3%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • 3DCodeBench

    Claude Sonnet 539.2%
    Source
    GPT-5.4 Pro—

    Not directly comparable

Reasoning

  • ARC-AGI-2

    Claude Sonnet 5—
    GPT-5.4 Pro83.3%
    Source

    Not directly comparable

Knowledge

  • HLE

    Claude Sonnet 557.4%
    Source
    GPT-5.4 Pro58.7%
    Source

    GPT-5.4 Pro leads this result

  • HLE w/o tools

    Claude Sonnet 543.2%
    Source
    GPT-5.4 Pro42.7%
    Source

    Claude Sonnet 5 leads this result

  • FrontierScience

    Claude Sonnet 5—
    GPT-5.4 Pro36.7%
    Source

    Not directly comparable

  • FrontierScience Research

    Claude Sonnet 5—
    GPT-5.4 Pro36.7%
    Source

    Not directly comparable

Math

  • IPhO 2025 (Theory)

    Claude Sonnet 5—
    GPT-5.4 Pro93.5%
    Source

    Not directly comparable

  • FrontierMath (legacy)

    Claude Sonnet 5—
    GPT-5.4 Pro50%
    Source

    Not directly comparable

  • FrontierMath v2 (Tiers 1-3)

    Claude Sonnet 5—
    GPT-5.4 Pro50.000%
    Source

    Not directly comparable

  • FrontierMath v2 (Tier 4)

    Claude Sonnet 5—
    GPT-5.4 Pro37.500%
    Source

    Not directly comparable

Multimodal

  • CharXiv

    Claude Sonnet 588.3%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • CharXiv w/o tools

    Claude Sonnet 577%
    Source
    GPT-5.4 Pro—

    Not directly comparable

  • MMMU-Pro

    Claude Sonnet 5—
    GPT-5.4 Pro94%
    Source

    Not directly comparable

Frequently asked questions

Which is better, Claude Sonnet 5 or GPT-5.4 Pro?

Claude Sonnet 5 has the higher public score estimate, 64.78 versus 61.26, 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 Sonnet 5 or GPT-5.4 Pro?

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, Claude Sonnet 5 or GPT-5.4 Pro?

The current agentic tasks averages use different weighted benchmark sets, so BenchLM does not name a winner from them. Read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Claude Sonnet 5 or GPT-5.4 Pro?

For the stated presets, chat costs $0.007 on Claude Sonnet 5 and $0.12 on GPT-5.4 Pro; repository review costs $0.13 and $2.04; the cache-heavy agent loop costs $0.18 and $8.40. GPT-5.4 Pro has no published cached-input rate, so cached tokens use its listed input rate.

Which has the larger context window, Claude Sonnet 5 or GPT-5.4 Pro?

GPT-5.4 Pro has the larger documented context window: 1.05M, compared with 1M.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 15, 2026

Watch Claude Sonnet 5 vs GPT-5.4 Pro

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

Read a sample issue

Join 2,000+ readers.