OpenAIproprietary
Released: 2024-11-20

GPT-4o

OpenAI versatile omni-model processing text, voice, audio, and visual inputs with low latency.

Intelligence Index
81.3/ 100
Calibrated multi-domain composite
Throughput Speed
70tok/sec
Output streaming throughput
Time To First Token
380ms
Initial chunk server latency
LiveBench Contam-Free
66.4%
September 2026 suite score

LiveBench Multi-Domain Evaluation Breakdown

Monthly refreshed contamination-free test sets
Logical Reasoning81.5%
Coding & Repo Repair66.5%
Mathematics (AIME)76.8%
Data Analysis & Tables78.4%
Language & Comprehension82.0%
Instruction Following90.8%

Sourced Benchmark Evaluations (13)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
StructFlowBenchNearing Saturation95%
75
2025-02-20independentSource ↗
Humanity's Last ExamActive3.3%
90
2025-01-24independentSource ↗
ARC-AGI-2Nearing Saturation63.6%
75
2024-12-18independentSource ↗
SWE-bench VerifiedSaturated50.5%
25
2024-11-29independentSource ↗
IFEvalDeprecated90.8%
25
2024-11-24independentSource ↗
SimpleQAActive38.2%
90
2024-11-07vendor-reportedSource ↗
HellaSwagSaturated95.3%
25
2024-08-01vendor-reportedSource ↗
ARC-AGISaturated5%
25
2024-06-27vendor-reportedSource ↗
AgentIFNearing Saturation58.5%
80
2024-06-01independentSource ↗
CommonsenseQASaturated85.4%
25
2024-05-13vendor-reportedSource ↗
GPQANearing Saturation56.1%
75
2024-05-13vendor-reportedSource ↗
MMLUSaturated88.7%
25
2024-05-13vendor-reportedSource ↗
MMLU-ProNearing Saturation72.6%
75
2024-05-13independentSource ↗
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
    temperature=0.2
)
print(response.choices[0].message.content)

Evaluation & Sourcing Notes

Scores listed for GPT-4o represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.