Skip to main content

GPT-5 (high)

OpenAIEstablishedReleased Aug 7, 2025
Overall Score
Est. 78Prov. #22 of 115
Arena Elo
1434
Categories Ranked
8of 8
Price (1M tokens)
$1.25 in / $10 out
Speed
83tok/s
Context
128K
ProprietaryReasoning
Confidence
reasoning

According to BenchLM.ai, GPT-5 (high) ranks #22 out of 115 models on the provisional leaderboard with an overall score of 78/100. It does not yet have enough sourced coverage for BenchLM's verified leaderboard. This places it in the mid-tier of AI models, with strengths in specific benchmark categories.

GPT-5 (high) is a proprietary model with a 128K token context window. It uses explicit chain-of-thought reasoning, which typically improves performance on math and complex reasoning tasks at the cost of higher latency and token usage.

GPT-5 (high) sits inside the GPT-5 family alongside GPT-5 (medium), GPT-5 mini, GPT-5 nano. This profile currently has 1 of 193 tracked benchmarks. BenchLM only exposes non-generated benchmark rows publicly, so missing categories stay blank until a sourced evaluation is available.

Its strongest category is Multimodal & Grounded (#7), while its weakest is Mathematics (#34). This performance profile makes it particularly strong for screenshots, documents, charts, and grounded multimodal workflows.

Ranking Distribution

Category rank across 8 benchmark categories — sorted by best rank

Category Performance

Scores across all benchmark categories (0-100 scale)

Category Breakdown

Agentic

#15
78.0/ 100
Weight: 22%0 benchmarks
Terminal-Bench 2.0BrowseCompOSWorld-VerifiedGAIATAU-benchWebArena

Coding

#33
71.8/ 100
Weight: 20%1 benchmark
SWE-bench VerifiedLiveCodeBenchSWE-bench ProSWE-RebenchSciCode

Reasoning

#17
76.7/ 100
Weight: 17%0 benchmarks
MuSRLongBench v2MRCRv2ARC-AGI-2

Knowledge

#18
80.3/ 100
Weight: 12%0 benchmarks
GPQASuperGPQAMMLU-ProHLEFrontierScienceSimpleQA

Math

#34
70.4/ 100
Weight: 5%0 benchmarks
AIME 2025BRUMO 2025MATH-500FrontierMath

Multilingual

#21
81.7/ 100
Weight: 7%0 benchmarks
MGSMMMLU-ProX

Multimodal

#7
92.3/ 100
Weight: 12%0 benchmarks
MMMU-ProOfficeQA ProCharXivCharXiv w/o tools

Inst. Following

#31
82.0/ 100
Weight: 5%0 benchmarks
IFEvalIFBench

Chatbot Arena Performance

Text Overall1434CI: ±4.531,971 votes
Coding1466CI: ±8.16,360 votes
Math1433CI: ±13.61,890 votes
Instruction Following1409CI: ±7.28,270 votes
Creative Writing1375CI: ±9.74,260 votes
Multi-turn1420CI: ±9.05,024 votes
Hard Prompts1446CI: ±5.914,989 votes
Hard Prompts (English)1453CI: ±7.57,681 votes
Longer Query1411CI: ±7.86,819 votes

Benchmark Details

Only benchmark rows with an attached exact-source record are shown here. Source-unverified manual rows and generated rows are hidden from model pages.

GPT-5 Family

reasoning · high

Frequently Asked Questions

How does GPT-5 (high) perform overall in AI benchmarks?

GPT-5 (high) currently ranks #22 out of 115 models on BenchLM's provisional leaderboard with an overall score of 78 (estimated). It is created by OpenAI and features a 128K context window.

Is GPT-5 (high) good for coding and programming?

GPT-5 (high) ranks #33 out of 115 models in coding and programming benchmarks with an average score of 71.8. There are stronger options in this category.

Which sibling models are related to GPT-5 (high)?

GPT-5 (high) belongs to the GPT-5 family. Related variants on BenchLM include GPT-5 (medium), GPT-5 mini, GPT-5 nano.

Does GPT-5 (high) have full benchmark coverage on BenchLM?

Not yet. GPT-5 (high) currently has 1 published benchmark scores out of the 193 benchmarks BenchLM tracks. BenchLM only exposes non-generated public benchmark rows, so missing categories stay blank until a sourced evaluation is available.

What is the context window size of GPT-5 (high)?

GPT-5 (high) has a context window of 128K, which determines how much text it can process in a single interaction.

Last updated: May 13, 2026 · Runtime metrics stay blank until BenchLM has a sourced snapshot.

Don't miss the next GPT moment

Which models moved up, what’s new, and what it costs. One email a week, 3-min read.

Free. One email per week.