aiquiz

Learn AI through quizzes & games

Start with the breakthrough quiz below—then test your knowledge of AI categories in the tabs, or try vocabulary, scramble, and the neural builder.

Breakthrough Quiz

Medium

More games

Stay Ahead of the Curve

Losing your sleep over AI?

Check these breakthroughs in time

hardware
High Impact

NVIDIA Blackwell Ultra B300

Next-gen GPU architecture with 208B transistors, delivering 2.5x performance improvement for transformer inference and 4x memory bandwidth.

GPUInferenceData Center
March 2026
NVIDIA GTC 2026
models
Critical Impact

GPT-5 Multimodal Reasoning

OpenAI's latest model demonstrates emergent mathematical reasoning capabilities and native multimodal understanding across text, images, audio, and video.

LLMMultimodalReasoning
February 2026
OpenAI Research
frontier
High Impact

Anthropic Constitutional AI v3

New alignment technique achieving 94% reduction in harmful outputs while maintaining helpfulness scores. Sets new benchmark for AI safety.

AlignmentSafetyRLHF
January 22, 2026
Anthropic Safety
Paradigm Shift
Critical Impact

Test-Time Compute Scaling

DeepMind proves that scaling compute during inference can match training-time scaling laws, fundamentally changing how we think about model capability.

Scaling LawsInferenceTheory
January 2026
DeepMind Research
hardware
High Impact

Cerebras WSE-3 Wafer Scale

Third generation wafer-scale engine with 4 trillion transistors, enabling training of 100T parameter models on a single system.

Wafer ScaleTrainingHPC
February 2026
Cerebras Systems
models
High Impact

Claude Opus 4.6 Extended Context

Anthropic’s Opus 4.6 brings a 1M token context window (beta), stronger coding and agentic work, and improved performance on long-horizon knowledge tasks—aligned with the official announcement.

Context LengthMemoryEnterprise
March 2026
Anthropic
frontier
Critical Impact

World Models for Robotics

Tesla AI demonstrates unified world model enabling zero-shot transfer of manipulation skills across different robot embodiments.

RoboticsWorld ModelsTransfer Learning
February 2026
Tesla AI Day
Paradigm Shift
High Impact

Liquid Neural Networks

MIT research shows liquid networks can achieve GPT-3 level performance with 1000x fewer parameters through continuous-time dynamics.

EfficiencyArchitectureNovel
January 2026
MIT CSAIL
hardware
High Impact

Photonic AI Accelerator

Lightmatter achieves 10 PFLOPS in 100W power envelope using photonic computing, promising 100x energy efficiency for inference workloads.

PhotonicsEnergy EfficiencyInference
March 2026
Lightmatter
Paradigm Shift
Critical Impact

Self-Improving Code Agents

Google DeepMind demonstrates AI systems that can recursively improve their own code, achieving superhuman performance on competitive programming.

Code GenerationSelf-ImprovementAGI
February 2026
DeepMind
frontier
Critical Impact

Mechanistic Interpretability Breakthrough

Researchers successfully reverse-engineer complete circuits in a 70B parameter model, enabling precise understanding of model behavior.

InterpretabilitySafetyResearch
March 2026
Anthropic Interpretability
models
High Impact

Gemini 3 Ultra Native Multimodality

Google's flagship model processes and generates interleaved text, images, audio, and video in a single unified architecture.

MultimodalGenerationUnified
January 2026
Google DeepMind

Frontier Labs and Research

A breakdown of leading labs, their flagship models, research, and unique contributions to the AI frontier.

Startup

San Francisco

OpenAI

Flagship Models:

  • GPT-5.4 family: Strong in math, business, multimodal, reasoning
  • o-series reasoning models: N/A
  • Sora: Video generation
  • Codex: Coding
  • Operator: Agent frameworks

Use Cases:

  • Consumer (ChatGPT for discovery/learning)
  • Enterprise (knowledge work, customer support, coding acceleration)
  • Education (math/science)
  • Agents (workflow automation)

Unique Proposition:

Democratized AI via ChatGPT/API; relentless scaling + reasoning breakthroughs; massive ecosystem and partnerships (e.g., Amazon, Oracle). Strong focus on safe deployment at scale.

Startup Journey/Funding:

Ultra-late stage / pre-IPO track. Raised ~$122B in a record February/March 2026 round (largest private AI deal ever). Valuation ~$850B+ post-money. Cumulative funding exceeds $180B+; projecting massive revenue but high capex burn.

Startup

San Francisco

Anthropic

Flagship Models:

  • Claude 4.6 family (Opus/Sonnet/Haiku): Leads in deep coding, agents, professional/enterprise tasks

Use Cases:

  • Coding/agents
  • Professional workflows
  • Scientific applications (e.g., NASA Perseverance rover autonomy)
  • Enterprise deployments

Unique Proposition:

Safety-first via Constitutional AI (models self-critique against principles); public benefit corporation structure; emphasis on alignment, harmlessness, and reliability over raw scale. Excels at "thinking" and complex tasks.

Startup Journey/Funding:

Late-stage powerhouse. $30B Series G (February 2026) at $380B post-money valuation. Total raised ~$64B+. Fast revenue growth (multi-billion ARR run-rate).

Startup

San Francisco / now tied to SpaceX

xAI

Flagship Models:

  • Grok 4: Multimodal, real-time capabilities
  • Grok Imagine: Video generation

Use Cases:

  • General exploration/universe understanding
  • API-powered apps (speed/multilingual)
  • Video generation
  • Enterprise tools
  • Real-time search/integration with X platform

Unique Proposition:

Maximum truth-seeking and curiosity (inspired by Hitchhiker’s Guide/Jarvis); no heavy censorship; real-time knowledge via X; now accelerated via SpaceX acquisition (Feb 2026) for hardware/AI synergy. Focus on understanding the universe.

Startup Journey/Funding:

Late-stage / acquired. $20B Series E (January 2026). Valuation $200B+ (pre-acquisition reports). Now integrated with SpaceX (targeting massive combined IPO valuation). Total raised ~$40B+.

Startup

Paris, France

Mistral AI

Flagship Models:

  • Mistral 3 family: Open-weights, efficient frontier models
  • Codestral: Specialized coding model

Use Cases:

  • Enterprise agents (orchestration, deep context)
  • Autonomous coding
  • Workflow automation
  • Content/search
  • Private/on-prem deployments (cloud/edge/devices)
  • Custom R&D

Unique Proposition:

Open-weights for customization/fine-tuning; privacy-first and sovereign (European); highly efficient/deployable models; full-stack platform with hands-on applied scientists. Balances frontier performance with control and cost.

Startup Journey/Funding:

Late-stage growth. Valuation $10B+ (2026 estimates). Multiple large rounds; focused on European AI independence. Not yet at OpenAI/Anthropic scale but gaining fast traction in enterprise.

Big Tech Lab

Multi-hub

Google DeepMind

Flagship Models:

  • Gemini 3 family: Multimodal, agentic, unified from ground up

Links:

Use Cases:

  • Integrated into Google products
  • Agentic/science benchmarks

Unique Proposition:

No VC funding rounds (part of Alphabet). Dominates in some agentic/science benchmarks.

Startup Journey/Funding:

Part of Alphabet

Big Tech Lab

Multi-hub

Meta AI

Flagship Models:

  • Llama 4: Open-weights, massive context windows like 10M tokens

Use Cases:

  • Democratizes access

Unique Proposition:

Public company; no startup valuation.

Startup Journey/Funding:

Part of Meta Platforms

Startup

San Francisco

World Labs

Flagship Models:

  • Marble: Multimodal world model for generating/editing persistent, physics-aware 3D environments from text/image/video/360 inputs

Use Cases:

  • Robotics simulation (motion/physics planning)
  • Architecture/design
  • AR/VR/gaming
  • Virtual production
  • Health systems modeling

Unique Proposition:

Spatial intelligence as the bridge from "seeing" to "doing/reasoning/creating" in 3D—frontier models for consistent, interactive world simulation that directly powers robotics and creative workflows.

Startup Journey/Funding:

Early-to-mid stage. ~$1B raised (2024–early 2026 rounds from NVIDIA, AMD, Autodesk, Fidelity, etc.); new funding announced Feb 2026. Valuation in $1B+ range. Focused on long-term foundational research with rapid product releases.

Startup

San Francisco

Physical Intelligence (π)

Flagship Models:

  • π0 / π0.5 / π*0.6 series: Generalist Vision-Language-Action (VLA) models
  • RL Token: Advancements
  • Multi-Scale Embodied Memory (MEM): Advancements
  • Real-time action chunking: Advancements

Use Cases:

  • Any robot / any task (dexterous manipulation, open-world mobile manipulators in kitchens/bedrooms, long-horizon tasks >10 min, human-to-robot video transfer)
  • Commercial via Physical Intelligence Layer partnerships

Unique Proposition:

"ChatGPT for robots"—general-purpose physical intelligence foundation models that generalize across hardware/tasks with minimal data, using experience-based RL and embodied memory.

Startup Journey/Funding:

Late-stage growth. Previously >$1B raised (Bezos, Khosla, Lux, Thrive, etc.); in talks for another ~$1B round (Mar 2026) at >$11B valuation (doubled from $5.6B in late 2025). Ex-DeepMind founders; ~80–100 employees scaling fast.

Startup

Paris / multi-hub

AMI Labs (Advanced Machine Intelligence)

Flagship Models:

  • World models based on JEPA architecture: Representation learning from video/spatial data for physics, memory, planning

Links:

Use Cases:

  • Robotics
  • Industrial automation
  • Healthcare—any domain needing physical reality modeling beyond language

Unique Proposition:

Alternative path to AGI via "world models" that learn latent structure/dynamics of reality (not just text prediction); persistent memory, reasoning, controllability, and safety.

Startup Journey/Funding:

Seed stage (launched late 2025/early 2026). Record $1.03B seed (Europe’s largest ever) at $3.5B pre-money valuation (Mar 2026). Backers: Bezos Expeditions, NVIDIA, Toyota Ventures, Samsung, Cathay, etc. Small team (~12 at launch) of top researchers.

Startup

New York / now integrated with Biohub

EvolutionaryScale

Flagship Models:

  • ESM3 (98B-parameter generative model): Reasoning over sequence/structure/function
  • ESM Cambrian family: Efficient representation learning

Use Cases:

  • De novo protein design (novel medicines, antibodies, carbon-capture enzymes, plastic-degrading enzymes, esmGFP fluorescent protein)
  • Programmable biology for scientists

Unique Proposition:

Frontier generative AI trained on 2.78B proteins / 771B tokens / 10²⁴ flops—enables "imagining" proteins via chain-of-thought prompting that evolution never produced. Open/safe/community-focused.

Startup Journey/Funding:

Early growth. $142M seed (2024); now joined Chan Zuckerberg Biohub for integrated AI + experimental biology push. Models in beta/commercial use.

Big Tech Lab

London; Google DeepMind spinout

Isomorphic Labs

Flagship Models:

  • Isomorphic Labs Drug Design Engine (IsoDDE): Unified AI system beyond AlphaFold 3 for structure prediction, binding affinity, pocket identification, and full drug design

Use Cases:

  • AI-powered drug discovery (protein-ligand binding, small-molecule design, novel therapeutics for diseases)
  • Accelerates from sequence to candidate medicines

Unique Proposition:

Bridges pure structure prediction to real-world drug design with step-change accuracy (2x+ over AlphaFold 3 on key benchmarks); digital-speed biology modeling.

Startup Journey/Funding:

Late-stage (spinout model). Backed by Alphabet/Google; no independent VC rounds disclosed—focus on pharma partnerships (e.g., J&J). Part of big-tech infrastructure but operates as dedicated drug-discovery AI lab.

Responsible AI lifecycle

Develop and deploy AI safely

Four phases—design, development, validation, and evaluation—with concrete product vendors teams use in safety-critical domains (not framework documents alone).

Where AI augments safety practice (illustrative)

Relative emphasis in recent tooling and org surveys—your mileage varies by domain.

Design → Evaluation

Design

Threat modeling, policy, and system boundaries before code ships.

  • AI copilots draft misuse scenarios, data-flow diagrams, and alignment specs from product briefs.
  • Constitutional and policy blocks are authored as structured artifacts—not afterthought prompts.
  • Data-classification and lineage tools map sensitive flows before models touch production data.

Product & platform vendors

Software teams ship these tools into regulated and safety-critical programs—not only generic AI ethics frameworks.

  • Credo AIBanking, insurance, healthcare AI governance

    Policy & risk register software for AI systems under audit.

  • IriusRiskFinancial services, medical devices, automotive suppliers

    Threat-modeling platform with libraries for regulated product teams.

  • ThreatModelerFinancial services, medical devices, automotive suppliers

    Automated threat modeling with compliance evidence for regulated software.

  • OneTrustHIPAA, GDPR, cross-border privacy programs

    Privacy & GRC workflows that feed into AI data-use decisions.

  • ServiceNow GRCEnterprise & public-sector compliance

    Control testing and evidence collection for AI change programs.

  • BigIDHealthcare, finance, energy

    Sensitive-data discovery and governance catalogs for training data scope.

  • CollibraHealthcare, finance, energy

    Enterprise data governance and lineage for AI readiness.

Development

Secure build pipelines, guardrails, and controlled training or fine-tuning.

  • IDE agents explain SAST/DAST findings and suggest minimal-risk patches with audit trails.
  • Synthetic and privacy-preserving data pipelines pair with confidential compute for sensitive fine-tunes.
  • Guardrail middleware, typed tools, and schema-first APIs ship next to model endpoints by default.

Product & platform vendors

Software teams ship these tools into regulated and safety-critical programs—not only generic AI ethics frameworks.

  • SnykSaaS, fintech, med-tech software

    SAST/SCA and dependency scanning in CI with policy gates.

  • SemgrepSecurity teams in regulated tech

    Custom static rules and supply-chain checks for application code.

  • VeracodeMedical device SW, automotive embedded, defense contractors

    Enterprise AppSec programs with release certification workflows.

  • CheckmarxMedical device SW, automotive embedded, defense contractors

    SAST, SCA, and DAST for enterprise AppSec with compliance reporting.

  • GitHub Advanced SecurityBroad enterprise; regulated via org policy

    Code scanning, secret scanning, and audit-friendly pull-request controls.

  • Protect AIML platform teams shipping models to production

    Model scanning and ML-BOM for third-party and in-house models.

  • HiddenLayerEnterprises deploying proprietary ML

    MLSOC tooling for model integrity and runtime anomaly signals.

Validation

Automated adversarial testing and benchmarks before release.

  • Large-scale red-teaming and jailbreak suites run in CI with multimodal and multilingual coverage.
  • Hybrid checks combine formal methods or property tests with LLM-based oracles for critical paths.
  • Safety benchmarks (e.g. domain-specific harm suites) gate merges like performance tests.

Product & platform vendors

Software teams ship these tools into regulated and safety-critical programs—not only generic AI ethics frameworks.

  • Robust IntelligenceFinance, healthcare, insurance LLM apps

    Continuous AI validation and firewall-style tests before deploy.

  • LakeraEnterprise LLM products

    Adversarial and jailbreak testing integrated into dev pipelines.

  • LatticeFlow AIEU-regulated industries, automotive perception teams

    Model validation and robustness analytics for high-stakes ML.

  • AporiaProduction ML teams in regulated sectors

    Custom monitors and validation hooks for model behavior pre-release.

  • PromptfooProduct engineering orgs shipping LLM features

    Eval harnesses and red-team suites runnable in CI.

Evaluation

Continuous monitoring, human oversight, and post-deployment assurance.

  • Production monitors track drift, refusal quality, tool misuse, and task failure—not only latency.
  • Human-in-the-loop review queues and escalation paths are instrumented for regulated deployments.
  • Post-market patterns (audit logs, incident playbooks, periodic conformity refresh) mirror safety-critical industries.

Product & platform vendors

Software teams ship these tools into regulated and safety-critical programs—not only generic AI ethics frameworks.

  • Arize AIFraud, credit, clinical decision support

    ML observability: drift, performance, and slice analysis in production.

  • WhyLabsIoT, healthcare analytics, industrial AI

    Data & model health monitoring with lightweight agents.

  • Fiddler AILending, hiring tools, regulated AI services

    Explainability, fairness metrics, and operational dashboards.

  • Arthur AICompliance-heavy ML in finance and HR tech

    Bias monitoring, compliance reporting, and alert routing.

  • Datadog AI ObservabilityMission-critical apps with ML components

    End-to-end observability tying model endpoints to system SLOs.

  • Dynatrace Davis AIMission-critical apps with ML components

    AIOps with causal AI for anomaly detection and incident routing.

Leading Companies

Grouped by scale—from global platforms to specialized players.

Terms of Use

Important Information & Disclaimer

The information provided on this website is for general informational and learning purposes only. It is compiled from publicly available sources and may, in some cases, be outdated.

**No Financial or Decision-Making Advice:** This site is not intended to provide financial, investment, legal, or any other professional advice. Do not use the information presented here for making critical decisions, financial planning, or any other high-stakes applications. The website is not responsible for any financial, mental, or addictive behaviors generated by using the quizzes or consuming the latest AI information presented herein.

**For Learning and Enjoyment:** We encourage you to use this resource for educational purposes, to expand your knowledge of the AI ecosystem, and to have fun exploring the various quizzes and content. Always consult with qualified professionals for specific advice tailored to your situation.

**Content Updates:** We aim to update the content on this site approximately once a month to keep it fresh and relevant.

While we strive to keep the information accurate and up-to-date, we make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability with respect to the website or the information, products, services, or related graphics contained on the website for any purpose. Any reliance you place on such information is therefore strictly at your own risk.

We reserve the right to modify these terms at any time. By using this website, you agree to these terms of use.