# General Compute

OpenAI-compatible inference API on heterogeneous compute

Source: https://wheredoirunthis.com/products/general-compute · JSON: https://wheredoirunthis.com/api/products/general-compute · Read on 2026-10-14

## At a glance

- Type: Inference cloud
- Kind: Hosted
- Isolation: Not disclosed
- Free tier: Not documented
- Starting price: $50/mo
- GPU: No
- Regions: 1
- Coverage: 6 of 61 assessed capabilities supported
- Agent readiness: read 10 of 11 signals, drive 9 of 11 signals

## Links

- Homepage: https://www.generalcompute.com/
- Docs: https://docs.generalcompute.com/
- Pricing: https://www.generalcompute.com/pricing

## Pricing

- Model: Usage-based inference; optional monthly plans; dedicated capacity/custom private model deployment separately quoted
- Unit: USD per million input/output tokens
- Free tier: $100 new-account API credit advertised on products/pricing; agent-signup docs still say $5

| Tier | Price | Includes | Notes |
| --- | --- | --- | --- |
| Pay As You Go | No monthly fee + token usage | 100 RPM; 200,000 TPM; 50,000 requests/day; 10,000,000 tokens/day; these are rate/day ceilings, not prepaid token allowances. | Human /pricing HTML: Priced per deployment, not per token; contact rate card. /pricing.md: self-serve usage inference + $100 credit. Docs publish token pricing and PAYG/Developer $50/mo/Scale $1000/mo/Enterprise custom. Preserve API SKU; pricing confidence medium due divergent surfaces. [source](https://docs.generalcompute.com/rate-limits.md) |
| Developer | $50/month + model token usage | 500 RPM; 1,000,000 TPM; 250,000 requests/day; 100,000,000 tokens/day; these are rate/day ceilings, not prepaid token allowances. | Human /pricing HTML: Priced per deployment, not per token; contact rate card. /pricing.md: self-serve usage inference + $100 credit. Docs publish token pricing and PAYG/Developer $50/mo/Scale $1000/mo/Enterprise custom. Preserve API SKU; pricing confidence medium due divergent surfaces. [source](https://docs.generalcompute.com/rate-limits.md) |
| Scale | $1,000/month + model token usage | 2,000 RPM; 5,000,000 TPM; 1,000,000 requests/day; 500,000,000 tokens/day; these are rate/day ceilings, not prepaid token allowances. | Human /pricing HTML: Priced per deployment, not per token; contact rate card. /pricing.md: self-serve usage inference + $100 credit. Docs publish token pricing and PAYG/Developer $50/mo/Scale $1000/mo/Enterprise custom. Preserve API SKU; pricing confidence medium due divergent surfaces. [source](https://docs.generalcompute.com/rate-limits.md) |
| Enterprise | Custom | Custom limits, dedicated infrastructure and SLAs; these are rate/day ceilings, not prepaid token allowances. | Human /pricing HTML: Priced per deployment, not per token; contact rate card. /pricing.md: self-serve usage inference + $100 credit. Docs publish token pricing and PAYG/Developer $50/mo/Scale $1000/mo/Enterprise custom. Preserve API SKU; pricing confidence medium due divergent surfaces. [source](https://docs.generalcompute.com/rate-limits.md) |

- Overages: PAYG no monthly minimum. MiniMax M2.7 $0.28/$1.20 input/output per 1M; DeepSeek V3.2 $0.25/$0.38; DeepSeek V3.1 and GPT-OSS120B $0.21/$0.79.
- Card required: Not documented
- Source: https://docs.generalcompute.com/models.md

## Website design

Desktop public website · assessed 2026-10-14

Complete assessment.

Dark, purple-accented marketing pages pair oversized white typography with a radiating hero graphic and outlined controls. The documentation switches to a bright, green-accented three-column reading layout.

- Styles: Corporate
- Landing page: captured
- Pricing: captured
- Docs: captured

## Runtime

- **GPU & accelerators**: No — "No rentable GPU instances in primary API SKU; GPU prefill exists in serving substrate." [source](https://www.generalcompute.com/.md) (confidence: medium)
- **BYOC (Bring your own cloud)**: na [source](https://www.generalcompute.com/products.md)
- **Secrets management**: na — "Bearer key management authenticates API access; it is not workload secrets management." (source: Inference API exposes no customer workload secrets storage/injection) (confidence: high)
- **Storage options**: na — "No customer workload storage" [source](https://www.generalcompute.com/products.md) (confidence: medium)
- **Supported workloads**: LLM inference, streaming chat, reasoning, tool calling, JSON output [source](https://docs.generalcompute.com/features.md)
- **Persistence**: Stateless prompt-to-tokens API [source](https://docs.generalcompute.com/api-reference/introduction.md)
- **Data jurisdiction**: United States; production us-west-2 — "New corporate markdown says existing US colocation; older infrastructure copy still promotes LATAM renewable sites. Do not carry Paraguay as current API location." [source](https://docs.generalcompute.com/index.md) (confidence: medium)
- **Resource limits & quotas**: PAYG 100 RPM / 200k TPM / 50k requests/day / 10M tokens/day; Developer 500 RPM / 1M TPM / 250k requests/day / 100M tokens/day; Scale 2000 RPM / 5M TPM / 1M requests/day / 500M tokens/day; Enterprise custom. [source](https://docs.generalcompute.com/rate-limits.md)
- **SDK languages**: Python, TypeScript, JavaScript [source](https://docs.generalcompute.com/quickstart.md)
- **Network & egress controls**: na — "Not applicable to inference API" [source](https://www.generalcompute.com/products.md) (confidence: medium)
- **Snapshots & forks**: na [source](https://www.generalcompute.com/products.md)
- **Deployment methods**: api, sdk [source](https://docs.generalcompute.com/quickstart.md)
- **Isolation**: Not disclosed — "API exposes no caller-code isolation boundary. Heterogeneous GPU prefill + purpose-built decode silicon substrate; no isolation primitive publicly guaranteed." [source](https://www.generalcompute.com/.md) (confidence: medium)

## Capabilities

### AI & agents

| Capability | Status | Evidence |
| --- | --- | --- |
| Agent skills | Supported | "Three published skills: agent-signup, openclaw-connect, opencode-connect." [source](https://www.generalcompute.com/.well-known/agent-skills/index.json) |
| AI app builder | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Autonomous QA agent | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Built-in coding agent | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| BYOK for LLMs | Not supported | "No advertised byok-llm capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| Credential-injection proxy | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| LLM gateway | Not supported | "No advertised llm-gateway capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| MCP client | Not supported | "No advertised mcp-client capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| MCP-server hosting | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| MCP server | Supported | "General Compute Docs MCP; documentation search/retrieval only. Corporate server-card says inference is REST, not MCP." [source](https://docs.generalcompute.com/.well-known/mcp/server-card.json) |
| Model inference API (first-party) | Supported | "All models are served via our OpenAI-compatible API." [source](https://docs.generalcompute.com/models.md) |
| Preinstalled coding agent | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Networking & delivery

| Capability | Status | Evidence |
| --- | --- | --- |
| Automatic HTTPS | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Custom domains | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Direct public IPv6 | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Edge KV store | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Egress controls | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Email send / receive | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Public preview URLs | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Private networking | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Built-in reverse proxy | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Organisation & security

| Capability | Status | Evidence |
| --- | --- | --- |
| Built-in auth | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| BYOC | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Deploy into customers' VPCs | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Secrets management | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Self-host / OSS core | Not supported | "No advertised self-host-oss capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| Sharing | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| SSO / SAML | Not supported | "No advertised sso-saml capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| Teams | Partially supported | "Organization-scoped model IDs and enterprise per-org allow lists; multi-user team account/member management not publicly documented." [source](https://docs.generalcompute.com/api-reference/endpoints/models-list.md) |
| Workload identity | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Developer access

| Capability | Status | Evidence |
| --- | --- | --- |
| CLI | Not supported | "No advertised cli capability in complete official API docs." [source](https://docs.generalcompute.com/llms-full.txt) |
| E2B-compatible API | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Filesystem API | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Git deploy | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| HTTP data API | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| OpenAPI spec | Supported | "openapi: 3.1.0; fetched JSON." [source](https://docs.generalcompute.com/api-reference/openapi.json) |
| Private container registry | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Public API | Supported | "Documented OpenAI-compatible inference endpoints." [source](https://docs.generalcompute.com/api-reference/introduction.md) |
| SDK | Supported | "Official Node.js/TypeScript and Python SDK installation documented." [source](https://docs.generalcompute.com/quickstart.md) |
| SSH access | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Compute & storage

| Capability | Status | Evidence |
| --- | --- | --- |
| Cloud-init / ignition config | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Desktop / computer-use sandbox | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| GPU compute | Not supported | "Prefill stays on GPUs; decode moves to purpose-built silicon. No GPU-backed customer instances provisioned by this self-serve API SKU. gpu remains fail for registry instance availability, not GPU-free substrate." [source](https://www.generalcompute.com/.md) |
| Live-state (memory) snapshots | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Multi-region | Partially supported | "All production traffic runs in us-west-2 today; additional/dedicated regions Enterprise-only." [source](https://docs.generalcompute.com/index.md) |
| Object storage | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Persistent disk | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Sandbox templates / custom images | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Scale to zero | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Selectable isolation runtime | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Snapshots / forking | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Data

| Capability | Status | Evidence |
| --- | --- | --- |
| Database branching | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Ephemeral instant DB (no-signup) | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| First-party database | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Vector search | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

### Platform & operations

| Capability | Status | Evidence |
| --- | --- | --- |
| Managed CI/CD runners | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Built-in observability | Partially supported | "API adds routing, observability, and ultra-low-latency infrastructure; no substantive metrics/logs/tracing interface documented." [source](https://docs.generalcompute.com/api-reference/introduction.md) |
| Preview environments | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Scheduled tasks | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Project template marketplace | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |
| Durable workflows | Not applicable | "Self-serve prompt-to-tokens API exposes no customer instance, app-hosting, workload networking or data-store surface; dedicated racks/root access are a separate sibling." [source](https://www.generalcompute.com/products.md) |

## Agent readiness

### Read: 10 of 11 signals present

10/11 signals (9 full, 2 partial)

- Llms txt: pass — "https://www.generalcompute.com/llms.txt — Substantive docs, pricing, API and integration links. Confidence: high."
- Llms full txt: pass — "https://docs.generalcompute.com/llms-full.txt — Complete public documentation; apex full text is 1,615,193 bytes and retained as POINTER. Confidence: high."
- Robots AI policy: pass — "https://www.generalcompute.com/robots.txt — Explicit AI crawler allow rules. Confidence: high."
- JSON LD: partial — "https://www.generalcompute.com/ — Organization, WebSite and ItemList; no Product/SoftwareApplication offers. Confidence: high."
- Sitemap: pass — "https://www.generalcompute.com/sitemap.xml — Non-trivial sitemap: 147 URLs. Confidence: high."
- Meta quality: pass — "https://www.generalcompute.com/ — Descriptive title, description and OpenGraph metadata. Confidence: high."
- Markdown negotiation: pass — "https://www.generalcompute.com/ — Accept: text/markdown returned HTTP 200 text/markdown; docs too. Confidence: high."
- Content efficiency: partial — "https://www.generalcompute.com/ — Readable text/HTML bytes = 1.6533% (script/style removed with HTMLParser). Confidence: high."
- Docs discoverable: pass — "https://www.generalcompute.com/ — Docs present in server-rendered primary navigation. Confidence: high."
- Pricing machine readable: pass — "https://docs.generalcompute.com/models.md — Structured token-price table and plan quota tables; corporate HTML pricing is now dedicated-deployment only. Confidence: high."
- Anti bot: pass — "https://www.generalcompute.com/ — All public marketing/docs/known-file requests were fetchable without challenge. Confidence: high."

### Drive: 9 of 11 signals present

9/11 signals (9 full, 0 partial)

- Public API: pass — "https://docs.generalcompute.com/api-reference/introduction.md — POST /v1/chat/completions, POST /v1/models/list, GET /v1/public/models documented. Confidence: high."
- Openapi spec: pass — "https://docs.generalcompute.com/api-reference/openapi.json — Fetched OpenAPI 3.1.0 JSON with paths, schemas, bearer authentication and OpenAIError. Confidence: high."
- MCP server: pass — "https://docs.generalcompute.com/.well-known/mcp/server-card.json — Documented Docs MCP: search and retrieve documentation; https://generalcompute.main-kill-isr.mintlify.me/mcp. NOT inference/control operations. Apex MCP descriptor conflicts with apex server-card explicitly stating no MCP server. Confidence: high."
- SDK packages: pass — "https://docs.generalcompute.com/quickstart.md — npm install @generalcompute/sdk; pip install generalcompute. Confidence: high."
- Auth docs: pass — "https://docs.generalcompute.com/api-keys.md — Bearer authentication, production base URLs, per-service keys, revocation and rotation. Confidence: high."
- Agent self serve creds: pass — "https://docs.generalcompute.com/agent-signup.md — Programmatic POST signup and verify return apiKey; requires user-provided email verification code. Documentation only, not exercised. Confidence: high."
- Webhooks: fail — "https://docs.generalcompute.com/llms-full.txt — No webhook integration/signing documentation in complete docs. Confidence: high."
- CLI tool: fail — "https://docs.generalcompute.com/quickstart.md — curl/examples are not an official CLI; none documented. Confidence: high."
- Error model: pass — "https://docs.generalcompute.com/api-reference/openapi.json — OpenAIError.error has message, type, param, code; 400/401 JSON responses documented. Confidence: high."
- Rate limit docs: pass — "https://docs.generalcompute.com/rate-limits.md — Plan RPM/TPM/day limits, 429 and x-ratelimit-limit/remaining/reset-requests headers. Confidence: high."
- Agent config: pass — "https://www.generalcompute.com/agents.md — agents.md plus published machine-readable agent-skills index and three installable SKILL.md documents. Confidence: high."

## Coding harnesses

- Claude Code (Anthropic) — "This works with any agent that can read your codebase and edit files: Claude Code, Codex, OpenCode, Cursor." [source](https://docs.generalcompute.com/migrate-with-ai.md)
- Codex CLI (OpenAI) — "This works with any agent that can read your codebase and edit files: Claude Code, Codex, OpenCode, Cursor." [source](https://docs.generalcompute.com/migrate-with-ai.md)
- OpenCode (opencode) — "This works with any agent that can read your codebase and edit files: Claude Code, Codex, OpenCode, Cursor." [source](https://docs.generalcompute.com/migrate-with-ai.md)
- Cursor (Anysphere) — "This works with any agent that can read your codebase and edit files: Claude Code, Codex, OpenCode, Cursor." [source](https://docs.generalcompute.com/migrate-with-ai.md)
- OpenClaw (Peter Steinberger) — "OpenClaw Integration: connect your OpenClaw AI assistant to General Compute." [source](https://docs.generalcompute.com/openclaw.md)

## Compliance & regions

- Regions: us-west-2
- Regions note: All production API traffic runs us-west-2; enterprise additional regions only. Current corporate deployment copy is US colo, conflicting older LATAM infrastructure copy.

## Company & presence

- Privacy policy: https://www.generalcompute.com/privacy
- Terms of service: https://www.generalcompute.com/terms
- GitHub: https://github.com/generalcompute
- Blog: https://www.generalcompute.com/blog
- Careers: https://www.generalcompute.com/careers

## Nearby in the atlas

- [Cerebrium](/products/cerebrium.md)

## Method

Facts come from the product's own public pages. Each claim carries its evidence: a quote and the page it came from. We count capabilities rather than grade products, and "Not assessed" means we have not checked, never that the product lacks it.