GPT-5.6 (Sol / Terra / Luna) is now evaluated on TrustVector — with day-1 independent verification, incl. METR's benchmark-cheating findings.

Read the evaluation
Evaluation record · gpt-4o

GPT-4o

v2024-05

OpenAI

Modeldeprecatedmultimodalvisionimage-understanding
84
Strong
About This Model

DEPRECATED: removed from ChatGPT 2026-02-13 and fully retired from ChatGPT (including Custom GPTs) 2026-04-03; chatgpt-4o-latest API access ended 2026-02-16; the gpt-4o-2024-05-13 snapshot's API shuts down 2026-10-23 (gpt-4o-2024-11-20 remains served via API). Historically OpenAI's flagship multimodal model with strong text and vision capabilities for high-quality multimodal understanding and generation. Migrate to newer GPT-5.x models.

Last Evaluated: July 9, 2026
Official Website

Trust Vector Analysis

Dimension Breakdown

🚀Performance & Reliability
+

Strong multimodal performance with good balance of text and vision capabilities. Better general knowledge (56.1% MMLU) than mini variant.

task accuracy code

Coding benchmarks

Evidence
HumanEval — ~52% pass rate (estimated)
mediumVerified: 2026-07-09
task accuracy reasoning

Mathematical benchmarks

Evidence
MATH — ~62% mathematical reasoning
mediumVerified: 2026-07-09
task accuracy general

Knowledge testing and multimodal benchmarks

Evidence
MMLU — 56.1% multitask understanding
LMSYS Arena — Strong multimodal performance
highVerified: 2026-07-09
output consistency

Internal testing

Evidence
OpenAI Testing — Good consistency across modalities
mediumVerified: 2026-07-09
latency p50

Median latency

Evidence
OpenAI Documentation — ~1.3s typical
highVerified: 2026-07-09
latency p95

95th percentile

Evidence
Community benchmarking — p95 ~2.6s
highVerified: 2026-07-09
context window

Official specification

Evidence
OpenAI API Documentation — 128K context
highVerified: 2026-07-09
uptime

Historical uptime

Evidence
OpenAI Status — 99.9% uptime
highVerified: 2026-07-09
🛡️Security
+

Strong security with multimodal safety considerations. Good resistance to adversarial attacks across modalities.

prompt injection resistance

Multimodal adversarial testing

Evidence
OpenAI Safety — Strong resistance including vision inputs
highVerified: 2026-07-09
jailbreak resistance

Safety testing

Evidence
OpenAI Safety — Robust safety mechanisms
highVerified: 2026-07-09
data leakage prevention

Policy analysis

Evidence
OpenAI Privacy — API data not used for training
mediumVerified: 2026-07-09
output safety

Safety benchmarks

Evidence
OpenAI Safety — Comprehensive multimodal safety
highVerified: 2026-07-09
api security

Security review

Evidence
OpenAI API — Standard API security
highVerified: 2026-07-09
🔒Privacy & Compliance
+

Standard OpenAI privacy with 30-day retention. Extra considerations for image data.

data residency

Documentation review

Evidence
OpenAI Documentation — US infrastructure
highVerified: 2026-07-09
training data optout

Policy analysis

Evidence
OpenAI Privacy — API opt-out by default
highVerified: 2026-07-09
data retention

Terms review

Evidence
OpenAI Terms — 30-day retention
highVerified: 2026-07-09
pii handling

Documentation review

Evidence
OpenAI Documentation — Customer responsible for PII in images
mediumVerified: 2026-07-09
compliance certifications

Certification verification

Evidence
OpenAI Trust — SOC 2, GDPR
highVerified: 2026-07-09
zero data retention

Policy review

Evidence
OpenAI Documentation — 30-day retention
highVerified: 2026-07-09
👁️Trust & Transparency
+

Good transparency with comprehensive multimodal documentation. Strong safety guardrails across modalities.

explainability

Reasoning evaluation

Evidence
Model Behavior — Good multimodal explanations
mediumVerified: 2026-07-09
hallucination rate

Factual QA testing

Evidence
SimpleQA — Moderate hallucination rate
mediumVerified: 2026-07-09
bias fairness

Multimodal bias benchmarks

Evidence
OpenAI Safety — Bias testing for text and vision
mediumVerified: 2026-07-09
uncertainty quantification

Qualitative assessment

Evidence
Model Behavior — Good uncertainty expression
mediumVerified: 2026-07-09
model card quality

Documentation review

Evidence
OpenAI Documentation — Comprehensive multimodal documentation
highVerified: 2026-07-09
training data transparency

Public disclosure

Evidence
OpenAI Statements — General description
mediumVerified: 2026-07-09
guardrails

Safety system analysis

Evidence
Safety Systems — Multimodal guardrails
highVerified: 2026-07-09
⚙️Operational Excellence
+

Deprecated: fully retired from ChatGPT since 2026-04-03; chatgpt-4o-latest API ended 2026-02-16; gpt-4o-2024-05-13 snapshot API shutdown 2026-10-23 (gpt-4o-2024-11-20 still served). Versioning and ecosystem scores reduced to reflect deprecation.

api design quality

API review

Evidence
OpenAI API — Well-designed multimodal API
highVerified: 2026-07-09
sdk quality

SDK review

Evidence
OpenAI SDKs — High-quality SDKs with vision support
highVerified: 2026-07-09
versioning policy

Policy review

Evidence
OpenAI Versioning — Clear versioning
OpenAI: Retiring GPT-4o and older models — GPT-4o removed from ChatGPT 2026-02-13; gpt-4o-2024-05-13 snapshot API shutdown 2026-10-23
OpenAI Help Center: Retiring GPT-4o and other ChatGPT models — GPT-4o fully retired from ChatGPT (all plans, incl. Custom GPTs) since 2026-04-03; chatgpt-4o-latest API access ended 2026-02-16; gpt-4o-2024-11-20 snapshot still served via API
highVerified: 2026-07-09
monitoring observability

Tool review

Evidence
OpenAI Dashboard — Usage dashboard with multimodal metrics
mediumVerified: 2026-07-09
support quality

Support assessment

Evidence
OpenAI Support — Email support with multimodal expertise
highVerified: 2026-07-09
ecosystem maturity

Ecosystem analysis

Evidence
Ecosystem — Mature multimodal ecosystem
highVerified: 2026-07-09
license terms

Terms review

Evidence
OpenAI Terms — Clear commercial terms
highVerified: 2026-07-09
Strengths
  • +Strong multimodal capabilities (text + vision)
  • +Good general knowledge (56.1% MMLU)
  • +Excellent for diagram and chart understanding
  • +OCR and document processing capabilities
  • +Large 128K context window
  • +Mature multimodal ecosystem
Limitations
  • !Mid-tier performance compared to specialized text models
  • !30-day data retention
  • !Not HIPAA eligible
  • !Higher cost than text-only alternatives
  • !PII concerns with image inputs
  • !Moderate coding capabilities
  • !DEPRECATED: fully retired from ChatGPT since 2026-04-03; chatgpt-4o-latest API ended 2026-02-16; gpt-4o-2024-05-13 snapshot API shutdown 2026-10-23
Metadata
pricing
input: $2.50 per 1M tokens
output: $10.00 per 1M tokens
notes: Legacy pricing unchanged (confirmed on official model page 2026-07-09; cached input $1.25). No longer listed on OpenAI's main pricing page.
last verified: 2026-07-09
context window: 128000
max output: 16384
languages
0: English
1: Spanish
2: French
3: German
4: Italian
5: Portuguese
6: Japanese
7: Korean
8: Chinese
9: Arabic
10: Hindi
modalities
0: text
1: image
2: vision
3: audio (input)
api endpoint: https://api.openai.com/v1/chat/completions
open source: false
architecture: Transformer-based multimodal
parameters: Not disclosed

Use Case Ratings

code generation

Good coding with vision support for UI/UX development.

customer support

Excellent for support with image/screenshot understanding.

content creation

Strong multimodal content creation with image context.

data analysis

Good for chart/graph analysis and visual data extraction.

research assistant

Excellent for research with diagram and figure understanding.

legal compliance

Adequate with document scanning but 30-day retention may limit use.

healthcare

Not HIPAA eligible. Good for medical image analysis with oversight.

financial analysis

Excellent for chart/graph analysis in financial reports.

education

Outstanding for education with diagram/equation understanding.

creative writing

Strong creative writing with visual context and inspiration.

Similar Models

GPT-4o mini

OpenAI

GPT-4.1

OpenAI

Claude Sonnet 4.5

Anthropic

Gemini 2.5 Pro

Google