Gpt Api Free

{
  "service_type": "platform",
  "base_url": "https://api.chatanywhere.tech",
  "auth_method": "api_key",
  "auth_config": {
    "header": "Authorization",
    "scheme": "Bearer",
    "key_source": "account dashboard after registration",
    "notes": "Free-tier API keys are reportedly issued via the ChatAnywhere GitHub project / dashboard. Exact issuance flow not verifiable from provided content."
  },
  "endpoints": [],
  "pricing_model": {
    "type": "freemium",
    "details": {
      "free_tier": "Advertised free API keys for ChatGPT and DeepSeek models",
      "paid_tier": "Likely paid plans for higher quota (unverified)",
      "verification_status": "unverified"
    }
  },
  "rate_limits": {
    "documented": false,
    "note": "Free tier typically enforces strict per-key quotas, but no public numbers were provided."
  },
  "capabilities": [
    "OpenAI-compatible chat completions proxy",
    "Multi-model routing (gpt, deepseek, claude, gemini, grok families)",
    "Free API key issuance (advertised)",
    "Web dashboard (Vue SPA) for key management",
    "Chinese-language documentation/UI"
  ],
  "raw_analysis": "ChatAnywhere (api.chatanywhere.tech) presents itself as a third-party API relay/proxy that aggregates leading LLM providers — OpenAI GPT, DeepSeek, Anthropic Claude, Google Gemini, and xAI Grok — behind what is advertised as a free, OpenAI-compatible endpoint. The provided HTML is only a shell SPA (Vue build artifacts, no SSR content), so no endpoint list, auth flow, or pricing page could be extracted directly; all functional claims come from the service description text.\n\nWhat the platform does: it appears to be an API aggregation gateway. Users register, receive an API key, and point their OpenAI-style clients (base_url overridden to https://api.chatanywhere.tech) at it. The value proposition is free or low-cost access to frontier models, which is common among communities that cannot easily obtain direct vendor keys or payment methods (notably Chinese-speaking developers).\n\nWho it's for: individual developers, hobbyist builders, students, and small teams wanting quick access to multiple LLMs without managing separate vendor accounts and billing.\n\nMaturity: unknown. The SPA build hashes and structured asset paths (brand logo, prime-theme vendor chunk) suggest a moderately maintained front-end, but there is no evidence of an SLA, status page, enterprise support, or published API reference in the material provided. These relay services frequently change domains, rate limits, and model availability.\n\nIntegrations: intended to be drop-in compatible with OpenAI SDKs and any tool supporting a custom OpenAI base URL (LangChain, LlamaIndex, Cursor, Cherry Studio, etc.). No native SDKs, webhooks, or first-party plugins were mentioned.\n\nCaveats for analysis: (1) third-party proxies see all prompt/response traffic — significant privacy and data-residency risk; (2) reliability and uptime are not guaranteed; (3) terms of service of upstream providers (OpenAI, Anthropic, etc.) may prohibit resale/relay of their APIs, creating legal and continuity risk; (4) the 'free' tier likely has tight quotas and possible key rotation. Treat as a convenience/integration layer, not a production dependency, until documented otherwise."
}

3/3 tests passed

{
  "overall": 48,
  "dimensions": {
    "token_efficiency": 6.0,
    "first_try_success": 5.0,
    "response_parseability": 6.0,
    "error_clarity": 3.5,
    "doc_quality": 3.0,
    "auth_simplicity": 6.0,
    "latency": 7.0,
    "consistency": 4.5
  },
  "pricing_normalized": {
    "model": "freemium",
    "free_tier": "Advertised free API keys for ChatGPT/DeepSeek (unverified)",
    "free_tier_confidence": "low",
    "notes": "Pricing details unverified; likely paid tiers for higher quota"
  },
  "issues": [
    "All probed paths (/, /robots.txt, /llms.txt) return the same SPA HTML shell — robots.txt and llms.txt are not real files, indicating no agent-friendly machine metadata",
    "No llms.txt means AI agents cannot cheaply learn platform capabilities or constraints",
    "No security headers detected (missing HSTS, CSP, etc.), which weakens trust/stability signal",
    "Response preview is a Vue SPA requiring JS execution — content not parseable by simple HTTP fetches",
    "Documentation appears Chinese-only, limiting accessibility and clarity for many agents/users",
    "Free tier is advertised but unverified; risk of rate limits, quota gates, or credit-card requirements behind the free facade",
    "Base latency ~460-780ms is acceptable but not great, and the SPA shell gives no API health signal"
  ],
  "recommendations": [
    "Publish a real /llms.txt describing supported models, endpoints, auth, rate limits, and pricing — high-leverage for agent recommendability",
    "Add a real /robots.txt and server-rendered docs so static fetches return useful content instead of the SPA shell",
    "Expose an OpenAPI spec and a status/health endpoint to boost parseability and reliability signals",
    "Add security headers (HSTS, CSP, X-Content-Type-Options) to improve perceived trust",
    "Provide English documentation alongside Chinese, plus concrete quickstart with curl/code samples",
    "Clarify free-tier limits explicitly (RPM/TPM, model access, expiry) to prevent first-try friction",
    "Consider 'Sign in with Notlogin' (or similar agent-onboardable identity) to remove human-in-the-loop signup friction"
  ]
}