Independent external evidence PRE-LAUNCH PREVIEW
AI Fast Now

External-first evidence · near-real-time source measurements and periodic public benchmarks. AI Fast Now direct measurements are historical; paid emergency verification is OFF by default.

AI status and model-route evidence

Official app and API status are separate from independent performance measurements.

AI Fast Now historical direct checks

Historical direct API checks, with their original timestamps and origins. Consumer apps may behave differently. How we measure.

OpenAI

AI Fast Now Direct Measurement · historical

  • No recorded direct evidence for this selection.

Paid direct verification is on operator-authorized emergency standby. Original model, origin and health-v1 workload remain separate from external routes.

Anthropic / Claude

AI Fast Now Direct Measurement · historical

  • No recorded direct evidence for this selection.

Paid direct verification is on operator-authorized emergency standby. Original model, origin and health-v1 workload remain separate from external routes.

Google Gemini

AI Fast Now Direct Measurement · historical

  • No recorded direct evidence for this selection.

Paid direct verification is on operator-authorized emergency standby. Original model, origin and health-v1 workload remain separate from external routes.

Normal paid measurement is paused; emergency verification requires operator authorization. No visitor action starts a provider call.

Benchmark / history

26 catalog models: 14 with near-real-time external evidence, 3 with periodic public benchmarks, 9 without a verified current external source. API measurements are separate from official status.

REAL OBSERVATIONS. CLEARER DECISIONS.

Explore ongoing model measurements

Independent model-route performance evidence.

  • Source, route and workload preserved
  • OpenAI, Claude and Gemini coverage
  • Best-time insights after sufficient history
Measurement source:Loading regional source…
Checking observationsLoading verified data…
Location & measurement details
Your location:Detecting approximate locationLocal time zone: checking…Auto preference

Loading regional observations…

Your AI Work Environment

—providers in benchmark data
—external measurement routes
—snapshot observations
OpenAI logoChatGPTChecking
Anthropic logoClaudeChecking
Google logoGeminiChecking

Checking verified public snapshots

A validated score needs more comparable observations.

Best Time to Work

Your local time zone

Outline marks your local time, not performance
Collecting
Collecting
Collecting
Collecting
Collecting
7-day evidence required

Distributed observations and held-out validation are required.

Model benchmark history

All models →
OpenAI logo

OpenAI / ChatGPT

7 catalog models · measured values only
External measurements · source / route / workload kept separate
ModelMonitoringLatest TTFTState / last measured
GPT-6 Astra——Collecting first measurement
GPT-6.1 Sol——Collecting first measurement
GPT-6 Luna——Collecting first measurement
GPT-6 SolPrevious generation——Collecting first measurement
GPT-5.6 SolPrevious generation——Collecting first measurement
GPT-5.6 TerraPrevious generation——Collecting first measurement
GPT-5.6 LunaPrevious generation——Collecting first measurement
View all OpenAI models →
Anthropic logo

Anthropic / Claude

8 catalog models · measured values only
External measurements · source / route / workload kept separate
ModelMonitoringLatest TTFTState / last measured
Claude Fable 5.1——Collecting first measurement
Claude Opus 5.5——Collecting first measurement
Claude Sonnet 5.5——Collecting first measurement
Claude Haiku 4.5——Collecting first measurement
Claude Fable 5Previous generation——Collecting first measurement
Claude Opus 5Previous generation——Collecting first measurement
Claude Sonnet 5Previous generation——Collecting first measurement
Claude Sonnet 4.5Previous generation——Collecting first measurement
View all Anthropic models →
Google logo

Google / Gemini

11 catalog models · measured values only
External measurements · source / route / workload kept separate
ModelMonitoringLatest TTFTState / last measured
Gemini 3.8 Flash——Collecting first measurement
Gemini 3.5 Flash-Lite——Collecting first measurement
Gemini 3.7 FlashPrevious generation——Collecting first measurement
Gemini 3.6 FlashPrevious generation——Collecting first measurement
Gemini 3.5 FlashPrevious generation——Collecting first measurement
Gemini 3.1 Flash-LitePrevious generation——Collecting first measurement
Gemini 2.5 ProPrevious generation——Collecting first measurement
Gemini 2.5 FlashPrevious generation——Collecting first measurement
Gemini 2.5 Flash-LitePrevious generation——Collecting first measurement
Gemini 3.1 Pro Preview——Collecting first measurement
Gemini 3 Flash Preview——Collecting first measurement
View all Google models →

Response Time Trend

OpenAI · selected model onlyLower TTFT means less waiting
Building the first performance timeline

Verified observations appear as they arrive. Gaps stay visible; missing hours are never estimated.

Individual observations · one source, route and workload · gaps remain visible. No interpolation or provider ranking.

Historical direct probe origins

All regions →

Historical direct readings · normal paid measurement paused

RegionOpenAI logoOpenAIAnthropic logoAnthropicGoogle logoGoogleState
N. VirginiaChecking snapshot———Checking
OregonChecking snapshot———Checking
SeoulChecking snapshot———Checking

Direct origins remain distinct from external provider routes. Coverage is not a claim about consumer app status.

Regional monitoringIndependent measurements
Real API observationsDirect provider endpoints
Global coverageRegional probe network
Developer focusedModels, metrics and methods

Monitoring activity

Coverage →

Verified snapshot refreshes. All times local.

  • Awaiting verified timestamps.
First report: dataset collection and review →

Why AI can feel slow

Possible factors, not a diagnosis of current conditions.

Network path

Distance and routing add travel time.

Model workload

Prompt length and reasoning affect work.

Service conditions

Rate limits or incidents can delay a request.

Understand response time →

Work smarter with AI

12 guides →
THE PERFORMANCE FIELD GUIDELatency basicsWhat time to first token really measuresLatency basicsWhy tokens per second can misleadInterpretationAPI performance is not ChatGPT app status

Methods and measurement limits for everyday AI work.

Check your connection to this website

Browser-to-site response time only. This does not call or test an AI model.