Skip to main content
BenchLM
Advertise

Model comparison

Claude Opus 5 vs GPT Realtime mini

Updated August 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Claude Opus 5

Anthropic

82.6/100

Supported · Public rank #3

90% interval 78.1–87.1

GPT Realtime mini

OpenAI

—

Evidence status unavailable

90% interval unavailable

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. Use the documented cost, context, and runtime rows instead.

0 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

    Claude Opus 5

    Claude Opus 5 has the larger documented context window.

    Confidence: documented

  • Chat turn cost

    1K fresh input + 500 output tokens

    GPT Realtime mini

    GPT Realtime mini 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

  • 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

  • Cache-heavy agent loop cost

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

    Not enough matched evidence

    The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. GPT Realtime mini does not fit this workload in one request.

    Confidence: listed-rates

  • Repository review cost

    50K fresh input + 3K output tokens

    Not enough matched evidence

    The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. GPT Realtime mini does not fit this workload in one request.

    Confidence: listed-rates

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
0
Claude Opus 5 only
63
GPT Realtime mini only
0
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
Claude Opus 5
90.8
GPT Realtime mini
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Coding

Not comparable
Claude Opus 5
89.5
GPT Realtime mini
Not measured
Weighted basis
2 vs 0 rows
Reading
Not comparable

Reasoning

Not comparable
Claude Opus 5
90.4
GPT Realtime mini
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Knowledge

Not comparable
Claude Opus 5
64.7
GPT Realtime mini
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Math

Not comparable
Claude Opus 5
Not measured
GPT Realtime mini
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multilingual

Not comparable
Claude Opus 5
Not measured
GPT Realtime mini
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multimodal

Not comparable
Claude Opus 5
66.9
GPT Realtime mini
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Instruction following

Not comparable
Claude Opus 5
Not measured
GPT Realtime mini
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

Claude Opus 5
$0.0175
Fits in one request
GPT Realtime mini
$0.0018
Fits in one request

GPT Realtime mini has the lower modeled cost

Costs use the listed standard API rates.

Repository review

50K fresh input + 3K output tokens

Claude Opus 5
$0.325
Fits in one request
GPT Realtime mini
$0.0372
Does not fit in one request

GPT Realtime mini does not fit this workload in one request.

Cache-heavy agent loop

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

Claude Opus 5
$0.45
Fits in one request
GPT Realtime mini
$0.048
Does not fit in one request

GPT Realtime mini does not fit this workload in one request.

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 5

$0.5 per 1M cached input tokens

Claude API pricing

GPT Realtime mini

$0.06 per 1M cached input tokens

OpenAI model documentation

Reasoning profile

Claude Opus 5

Reasoning

GPT Realtime mini

Non-Reasoning

Weight access

Claude Opus 5

Proprietary

GPT Realtime mini

Proprietary

License

Claude Opus 5

Proprietary

GPT Realtime mini

Proprietary

Release date

Claude Opus 5

2026-07-24

GPT Realtime mini

Not sourced

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
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict.
Workload cost
Repository review: $0.325 vs $0.0372. Cache-heavy agent loop: $0.45 vs $0.048.
Context tradeoff
Claude Opus 5 has the larger documented window (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 evidence63 rows

Agentic

  • FrontierBench v0.1

    Claude Opus 543.3%
    Source
    GPT Realtime mini—

    Not directly comparable

  • BrowseComp

    Claude Opus 590.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HLE w/ tools

    Claude Opus 564.7%
    Source
    GPT Realtime mini—

    Not directly comparable

  • DeepSearchQA

    Claude Opus 595.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • DRACO

    Claude Opus 588.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • BrowseComp (10-agent, prerelease)

    Claude Opus 593.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • OSWorld 2.0

    Claude Opus 570.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • MCP Atlas

    Claude Opus 585.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • MCP-Atlas claim coverage

    Claude Opus 589.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • LAB all-pass (Anthropic harness)

    Claude Opus 523.58%
    Source
    GPT Realtime mini—

    Not directly comparable

  • LAB criterion-pass (Anthropic harness)

    Claude Opus 593.74%
    Source
    GPT Realtime mini—

    Not directly comparable

  • LAB all-pass (Harvey held-out)

    Claude Opus 511.7%
    Source
    GPT Realtime mini—

    Not directly comparable

  • LAB criterion-pass (Harvey held-out)

    Claude Opus 594.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Toolathlon-Verified

    Claude Opus 580.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Toolathlon Verified Pass@3

    Claude Opus 587.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Toolathlon Verified Pass³

    Claude Opus 573.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Toolathlon Verified avg. turns

    Claude Opus 523.5 turns
    Source
    GPT Realtime mini—

    Not directly comparable

  • AutomationBench

    Claude Opus 526.0%
    Source
    GPT Realtime mini—

    Not directly comparable

Coding

  • SWE-bench Verified

    Claude Opus 596%
    Source
    GPT Realtime mini—

    Not directly comparable

  • SWE-bench Pro

    Claude Opus 579.2%
    Source
    GPT Realtime mini—

    Not directly comparable

  • SWE Multilingual

    Claude Opus 589.5%
    Source
    GPT Realtime mini—

    Not directly comparable

  • SWE Multimodal

    Claude Opus 559.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • deepSwe

    Claude Opus 568.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • FrontierCode 1.1 Main

    Claude Opus 553.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • FrontierCode 1.1 Extended

    Claude Opus 563.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ProgramBench (episode 1)

    Claude Opus 583.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ProgramBench

    Claude Opus 593.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • cursorBench32

    Claude Opus 570.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • VulcanBench v3

    Claude Opus 587.0%
    Source
    GPT Realtime mini—

    Not directly comparable

Reasoning

  • ARC-AGI-1

    Claude Opus 597.50%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ARC-AGI-2

    Claude Opus 590.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ARC-AGI-3

    Claude Opus 530.2%
    Source
    GPT Realtime mini—

    Not directly comparable

Knowledge

  • HLE

    Claude Opus 564.7%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HLE w/o tools

    Claude Opus 556.3%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HealthBench (raw)

    Claude Opus 567.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HealthBench (length-adjusted)

    Claude Opus 557.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HealthBench Professional

    Claude Opus 559.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • HealthBench Professional (raw)

    Claude Opus 573.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • BioMysteryBench (human-solvable)

    Claude Opus 590.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • BioMysteryBench (human-difficult)

    Claude Opus 549.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • SpatialBench Verified

    Claude Opus 572.5%
    Source
    GPT Realtime mini—

    Not directly comparable

  • SingleCellBench

    Claude Opus 560.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ProteinGym Hard

    Claude Opus 547.7%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Protein Design

    Claude Opus 542.5%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Organic chemistry V2

    Claude Opus 561.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Protocols (troubleshooting)

    Claude Opus 561.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Protocols (understanding)

    Claude Opus 578.4%
    Source
    GPT Realtime mini—

    Not directly comparable

Math

  • IMO 2026

    Claude Opus 542/42
    Source
    GPT Realtime mini—

    Not directly comparable

  • RiemannBench (no tools)

    Claude Opus 560.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • RiemannBench (tools)

    Claude Opus 579.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ArXivMath Jun. 2026 (no tools)

    Claude Opus 590.8%
    Source
    GPT Realtime mini—

    Not directly comparable

  • ArXivMath Jun. 2026 (tools)

    Claude Opus 591.3%
    Source
    GPT Realtime mini—

    Not directly comparable

Multilingual

  • GMMLU

    Claude Opus 592.5%
    Source
    GPT Realtime mini—

    Not directly comparable

  • MILU

    Claude Opus 592.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • INCLUDE

    Claude Opus 589.8%
    Source
    GPT Realtime mini—

    Not directly comparable

Multimodal

  • Chartography (no tools)

    Claude Opus 529.6%
    Source
    GPT Realtime mini—

    Not directly comparable

  • Chartography (tools)

    Claude Opus 583.0%
    Source
    GPT Realtime mini—

    Not directly comparable

  • BenchCAD Vision2Code (no tools)

    Claude Opus 50.366
    Source
    GPT Realtime mini—

    Not directly comparable

  • BenchCAD Vision2Code (tools)

    Claude Opus 50.821
    Source
    GPT Realtime mini—

    Not directly comparable

  • GDP.pdf (no tools)

    Claude Opus 583.4%
    Source
    GPT Realtime mini—

    Not directly comparable

  • GDP.pdf (tools)

    Claude Opus 585.5%
    Source
    GPT Realtime mini—

    Not directly comparable

  • OfficeQA

    Claude Opus 578.1%
    Source
    GPT Realtime mini—

    Not directly comparable

  • OfficeQA Pro

    Claude Opus 566.9%
    Source
    GPT Realtime mini—

    Not directly comparable

Frequently asked questions

Which is better, Claude Opus 5 or GPT Realtime mini?

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Claude Opus 5 or GPT Realtime mini?

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 Opus 5 or GPT Realtime mini?

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, Claude Opus 5 or GPT Realtime mini?

For the stated presets, chat costs $0.0175 on Claude Opus 5 and $0.0018 on GPT Realtime mini; repository review costs $0.325 and $0.0372; the cache-heavy agent loop costs $0.45 and $0.048. GPT Realtime mini does not fit this workload in one request.

Which has the larger context window, Claude Opus 5 or GPT Realtime mini?

Claude Opus 5 has the larger documented context window: 1M, compared with 32K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 4, 2026

Watch Claude Opus 5 vs GPT Realtime mini

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

Read a sample issue

Join 2,000+ readers.