Benchmark

OpenAI — reported benchmarks

OpenAI · source page ↗ · last checked Aug 22, 2026, 6:02 PM

Reported benchmarks · GPT-5.6 Sol

captured Aug 22, 2026, 6:02 PM

Benchmark	Score
Terminal-Bench 2.1	88.8% · base Sol; Sol Ultra scored 91.9%
Agents' Last Exam	53.6 score · max reasoning setting
ARC-AGI-3	38.3% · with retained reasoning and compaction enabled
ARC-AGI-3	13.3% · official harness (standard)
DeepSWE	72.7%
BrowseComp	90.4% · base Sol; Sol Ultra scored 92.2%
ExploitBench	73.5%
Capture-the-Flag	96.7% · OpenAI's curated internal task set; tool-enabled harness
HealthBench Professional	60.5 length-adjusted score

Vendor-reported via automated web search — not independently verified. See the cited matrix on /models.

In the news · OpenAI

OpenAI Cuts GPT-5.6 Sol API Pricing 20% for Three Months via citybiz — OpenAI Cuts GPT-5.6 Sol API Pricing 20% for Three Months citybiz · Aug 24, 2026, 9:46 AM
OpenAI Launches ChatGPT for Teens With Study-Focused Features and Default Safety Protections via gHacks — OpenAI Launches ChatGPT for Teens With Study-Focused Features and Default Safety Protections gHacks · Aug 24, 2026, 8:27 AM
Longbridge Launches on ChatGPT and Claude via Yahoo Finance Singapore — Longbridge Launches on ChatGPT and Claude Yahoo Finance Singapore · Aug 24, 2026, 8:00 AM
AI Search Reshapes Demand as ChatGPT Pre-Priced Buyers via MarTech Cube — AI Search Reshapes Demand as ChatGPT Pre-Priced Buyers MarTech Cube · Aug 24, 2026, 6:37 AM
Nvidia Is Coming for OpenAI and DeepSeek With $6 Billion Poolside Deal to Build a Powerful Open-Weight AI via Benzinga — Nvidia Is Coming for OpenAI and DeepSeek With $6 Billion Poolside Deal to Build a Powerful Open-Weight AI Benzinga · Aug 24, 2026, 4:21 AM
Remember Hugging Face? The AI Hub Hit by an OpenAI Security Breach Is Reportedly Exploring a $13 Billion Sale via Benzinga — Remember Hugging Face? The AI Hub Hit by an OpenAI Security Breach Is Reportedly Exploring a $13 Billion Sale Benzinga · Aug 24, 2026, 3:00 AM
OpenAI Unveils Harness Post-Black Whale Launch: Competing for Dominance in the Agent Runtime Space via 36Kr — OpenAI Unveils Harness Post-Black Whale Launch: Competing for Dominance in the Agent Runtime Space 36Kr · Aug 24, 2026, 2:33 AM
OpenAI Acquires 10,000-Star GitHub Agent "AI Version of Firebase": Entire Team Officially Joins OpenAI via 36Kr — OpenAI Acquires 10,000-Star GitHub Agent "AI Version of Firebase": Entire Team Officially Joins OpenAI 36Kr · Aug 24, 2026, 1:43 AM

Importance-filtered press coverage (Google News) mentioning OpenAI. Headlines link to the original; verify before acting.

Change history

Vendor claim Aug 22, 2026, 6:02 PM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 9 benchmark claims (via web search)
Vendor claim Aug 15, 2026, 12:02 PM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 11 benchmark claims (via web search)
Vendor claim Aug 8, 2026, 12:02 AM

OpenAI reported benchmarks updated

GPT-5: 6 benchmark claims (via web search)
Vendor claim Aug 1, 2026, 12:02 AM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 11 benchmark claims (via web search)
Vendor claim Jul 24, 2026, 6:02 PM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 5 benchmark claims (via web search)
Vendor claim Jul 17, 2026, 12:02 PM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 12 benchmark claims (via web search)
Vendor claim Jul 10, 2026, 6:02 AM

OpenAI reported benchmarks updated

GPT-5.6 Sol: 9 benchmark claims (via web search)
Vendor claim Jul 2, 2026, 6:02 PM

OpenAI reported benchmarks updated

GPT-5.5: 12 benchmark claims (via web search)
Vendor claim Jun 25, 2026, 1:19 PM

OpenAI reported benchmarks updated

GPT-5.5: 2 benchmark claims (via web search)