MidnightAI.org
Tracking humanity's progress toward superintelligent AI
Live Analysis
4
Minutes to Midnight
Confidence
75%
Current Analysis
“The developments show steady infrastructure progress rather than capability breakthroughs. The shift to custom ASICs and China's alternative chip architectures suggest the field is diversifying approaches rather than achieving transformative advances.”
Key Factors
- Custom AI chip development accelerating with ASICs projected to outpace GPUs by 2026
- Growing focus on agentic AI systems but implementation challenges remain significant
- China's 'overtaking on curves' strategy showing results in quantum computing and chip design
AI Capabilities
100→
🧠Reasoning
88→
💬Language
87→
💻Coding
86→
👁️Multimodal
68→
🤖Agency
61→
🔬Science
50→
🦾Robotics
Weighted Average80%
Latest AI News
Unrolling the Codex agent loop
A technical deep dive into the Codex agent loop, explaining how Codex CLI orchestrates models, tools, prompts, and performance using the Responses API.
blog
Personal Intelligence in AI Mode in Search: Help that's uniquely yours
social
AI Mania Is Eviscerating Global Decision-Making
social
Mayor Mamdani Says Landlords Can't Use AI Images to Advertise
social
Setting up your spare Mac for Claude Code to control, a step-by-step guide
social
Our Approach to Bioresilience: Isomorphic Labs and Google DeepMind
social
Show HN: Q3Edit – Edit and play Quake 3 maps in the browser
30-Day Trend
No historical data yet. Predictions will appear here as they accumulate.
AI Milestones
View All →AI Milestones
Autonomous Research Agent
PredictedDec 2026In 4mo
Humanoid Robot Mass Production
PredictedJun 2027In 11mo
Claude Discovers macOS Kernel Vulnerability CVE-2026-28952
AnthropicMay 20262mo ago
OpenAI Prepares for IPO Filing
OpenAIMay 20262mo ago
Anthropic Expands to Colossus2 with GB200 GPUs
AnthropicMay 20262mo ago
AI Companies
AI21 Labs
Jamba 1.5, Jurassic-2
Avg. capability: 48%
Aleph Alpha
Luminous, Pharia
Avg. capability: 51%
Alibaba (Qwen)
Qwen-2.5 Max, Qwen-VL
Avg. capability: 58%
Anthropic
Claude 3.5 Sonnet, Claude 3 Opus
Avg. capability: 64%
Baidu (Ernie)
Ernie 4.0, Ernie Bot
Avg. capability: 63%
ByteDance (Doubao)
Doubao 1.5 Pro, Seed series
Avg. capability: 56%
Cohere
Command R+, Embed v3
Avg. capability: 54%
DeepSeek
DeepSeek-V3, DeepSeek-R1
Avg. capability: 57%
Google DeepMind
Gemini 2.0, Gemini 1.5 Pro
Avg. capability: 70%
Meta AI
Llama 3.1, Llama 3.2
Avg. capability: 55%
Microsoft AI
Copilot, Phi-3
Avg. capability: 60%
Mistral AI
Mistral Large 2, Mixtral 8x22B
Avg. capability: 57%
OpenAI
GPT-4o, GPT-4 Turbo
Avg. capability: 68%
xAI
Grok-2, Grok-1
Avg. capability: 49%
Compare
Side-by-side analysis