01 / The problem AI labs face
Your model launches. A user in Country X can't load it.
You don't know if it's a network issue, an ISP block, or a regulatory move. Your support team writes a generic “sorry, working on it” reply. Your policy team has no data to ground a decision. Your product team has to guess at rollout priorities. Your competitor's service may already be working in that country.
Voidly answers, in one API call, whether your service is accessible from inside that country's network — and how it's being blocked if not.
02 / The matrix (live preview)
AI services × restrictive countries.
Each cell reflects either provider-stated restriction (yellow) or network-confirmed blocking from our probes (red). Refreshed every 5 minutes.
| Country | ChatGPT | Claude | Gemini | Mistral | DeepSeek | Meta AI | Grok | Perplexity |
|---|---|---|---|---|---|---|---|---|
| CNChina | ||||||||
| IRIran | ||||||||
| RURussia | ||||||||
| KPNorth Korea | ||||||||
| CUCuba | ||||||||
| SYSyria | ||||||||
| BYBelarus | ||||||||
| VEVenezuela | ||||||||
| TRTurkey | ||||||||
| AEUAE | ||||||||
| PKPakistan | ||||||||
| KZKazakhstan |
03 / The asymmetry data
Your service is not your competitor's service.
Each AI lab has a distinct accessibility footprint. This is the highest-leverage data we sell — every cell here is a commercial decision someone on your team should be able to make.
Claude reaches places ChatGPT and Gemini do not — a competitive asymmetry your policy and rollout teams should be tracking week-over-week.
ChatGPT is the most-searched-for chat assistant globally, and also the most blocked. Network methods, country list, and the exact ASNs blocking you — we have the data.
Gemini is the most-restricted of the major Western chat assistants in our data. The breakdown of provider-restricted vs network-blocked is the conversation worth having.
04 / What we offer
Six things only Voidly does for AI labs.
Real-time accessibility monitoring
Signed webhook fires when your service's status changes in a country we monitor — DNS poisoning, TCP reset, blockpage, TLS interference, or full unreachability. HMAC-SHA256 signed, retry with exponential backoff (2/4/8/16/32 min). Per-country, per-domain, per-method filters.
Pre-shipping diligence
Before you launch a new model surface or domain, we test it from 130 countries on demand. Returns accessibility status, observed methods, and any historical incidents involving the domain or related infra. One API call, machine-readable output.
Regulatory mapping
We separate provider-stated restrictions (your own ToS / supported-country list) from network-level blocking (DPI, DNS poisoning, BGP withdrawal). Your policy team gets a clean dashboard of where you have chosen not to ship vs where you are being blocked.
Competitive intelligence
How does your service's accessibility compare with peer labs? Side-by-side, country-by-country, method-by-method. The asymmetry is the value — Claude is accessible in places ChatGPT is blocked, and vice versa. We track this longitudinally.
Custom probe deployment
We already operate 37+ globally distributed nodes. Enterprise customers can request additional probes in target jurisdictions — physical hosting, AS-level placement, or your own VM. Lead time depends on the region, typically 2-6 weeks.
MCP integration for self-monitoring
Your agents can self-monitor their own service status via @voidly/mcp-server (83 tools, on npm). Use cases: automatically failover to a peer service when blocked; notify support before the user does; gate features by accessibility tier per region.
05 / Concrete use cases
Three scenarios you have right now.
Regional rollout verification
You launch your model in a new region. Voidly verifies accessibility from probes physically inside that region — DNS resolution, TCP handshake, TLS, end-to-end loadability of the chat surface. Before your support team learns about a problem from social media.
Policy-team compliance answer
Your policy team needs to answer "is our service accessible in country X" for a regulator, a partnership conversation, or a data-export questionnaire. Voidly returns the answer — with citation-grade evidence — in one API call. No vague "we think so."
Comparative blocking telemetry
Your release team ships a new model surface (new domain, new subdomain, new region) and wants to know if it gets blocked faster than your standard product. Voidly tracks every domain you point us at, longitudinally, with method-level diagnostics.
06 / Proof
Verifiable now. No paywall on the data.
The whole stack is open before you talk to us. Every claim on this page is measurable from public endpoints.
# Is ChatGPT accessible in China?
curl "https://api.voidly.ai/v1/accessibility/check\
?domain=chat.openai.com&country=CN"
# Same for Claude — note the asymmetry
curl "https://api.voidly.ai/v1/accessibility/check\
?domain=claude.ai&country=CN"
# Batch — every major AI service in one call
curl -X POST https://api.voidly.ai/v1/accessibility/batch \
-H "Content-Type: application/json" \
-d '{"domains":["chat.openai.com","claude.ai","gemini.google.com",
"chat.mistral.ai","x.ai","meta.ai"],
"country":"RU"}'07 / Pricing
Enterprise data feeds — schedule a call.
The public API is free, CC BY 4.0, and rate-limited. AI lab feeds are priced per use case — number of monitored domains, alert webhook volume, custom-probe deployments, embargo access, white-label requirements.
We don't list a sticker price because we haven't commoditized this yet and refuse to overcharge a team whose use case doesn't need the full stack. A 30-min call gets you a fixed quote and a pilot start date.
Most AI-lab pilots start within a week of the discovery call.
08 / By name
We are watching you. Politely.
These are statements of what our public data shows, not endorsements or partnerships. If anything below is wrong, tell us — we will publish a correction with the same prominence.
09 / Talk to us
Pilot in a week. Production in two.
One email gets you a 30-min discovery call, a sandbox API key, and sample integration code (LangChain, CrewAI, raw HTTP, your stack).