{"slug": "private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable", "title": "Private, OpenAI-compatible LLM APIs compared: zero retention, TEE and verifiable enclaves", "summary": "A developer at Morse published a comparison of seven OpenAI-compatible LLM APIs—Morse, OpenRouter, TrustedRouter, Phala/RedPill, NEAR AI Cloud, Tinfoil, Prem and Privatemode—on zero-retention, TEE and end-to-end encryption, and verifiable attestation, checked against each vendor's own docs as of November 20, 2026. The writeup distinguishes contract-based zero retention, hardware enclaves and attestation reports signed by the chip vendor, and notes that Morse's enclave code is open source and checkable with `npx @askmorse/verify`. It flags that OpenRouter's self-serve accounts allow safety-retention endpoints by default unless ZDR is requested per call.", "body_md": "*Every fact about each API comes from its own docs, linked inline, checked on November 20, 2026.*\n\nIf you send prompts to a hosted model, \"private\" can mean any of three different things:\n\n**Attestation** ties it together. It's a report signed by the hardware vendor that says what code an enclave is running, so you can check it against published source instead of taking the provider's word for it.\n\nMorse first (it's ours), then the others by type: routers, enclave hosts, gateways. This isn't a ranking.\n\n| API | OpenAI-compatible | ZDR / no retention | TEE | E2EE | Attestation you can check | Models | \n|---|---|---|---|---|---|---|\n| Morse (ours) | Yes | Yes, `private` tier by contract; Morse's enclave keeps no prompt or reply | Yes, `tee` tier | Yes, `e2ee` tier, to the model's enclave | Yes, `npx @askmorse/verify` | 87 open-weight models | \n| OpenRouter | Yes | Yes, per request ( `zdr: true` ) | Not described on its pages | Not described | Not described | Open and proprietary | \n| TrustedRouter | Yes | Yes (prompt storage off) | Gateway in a TEE; confidential routes via other hosts | TLS ends inside the gateway's enclave | Yes, image digest at `/attestation` | Open and proprietary | \n| Phala / RedPill | Yes | Gateway stores no bodies; `zdr: true` upstream | Yes (Intel TDX, NVIDIA H100/H200/B300) | To the gateway | Yes, `@phala/aci-verifier` , dstack verifier | Open and proprietary (proprietary outside the TEE) | \n| NEAR AI Cloud | Yes | \"No logs of prompts or responses\" | Yes (Intel TDX + NVIDIA H200) | To the model's enclave | Yes, open-source verifier + web page | Open in TEE; GPT/Claude/Gemini outside it | \n| Tinfoil | Yes | Yes | Yes (AMD SEV-SNP, Intel TDX, NVIDIA H100/H200/B200) | To the model's enclave (TLS in enclave + EHBP) | Yes, SDKs + `tinfoil attestation verify` | Open-weight only | \n| Prem | Yes (gateway or local proxy) | Yes, in ZDR mode | Yes, in confidential mode (SEV-SNP/TDX + NVIDIA) | Yes, in confidential mode (ML-KEM768 + X25519) | Yes, Reticle verifier | Open-weight only | \n| Privatemode | Yes (local proxy) | Prompts not stored | Yes (SEV-SNP, TDX, NVIDIA H100/B200) | To the worker (HPKE) | Yes, via the proxy; reference values public | Open-weight only | \n\n\"Not described\" means we couldn't find it on that company's own pages. It doesn't mean the feature doesn't exist.\n\n**Morse (ours).** One OpenAI-compatible endpoint for 80+ open-weight models, \"called from Morse's attested enclave, which keeps no prompt or reply\" ([docs](https://trymorse.ai/inference/docs?src=seo-devto)). TLS terminates inside the enclave, and providers see Morse as the customer, so \"your account, your card, and your IP address stop at Morse\" ([inference](https://trymorse.ai/inference?src=seo-devto)). You set a privacy floor per request: `private` (providers bound by contract to keep nothing), `tee` (the model runs in the provider's enclave) or `e2ee` (encrypted to a key only the model's enclave holds). Morse \"refuses a request it can't serve at that floor rather than sending it lower.\" The enclave code is [open source](https://github.com/trymorse/morse-enclave), and `npx @askmorse/verify inference.trymorse.ai` checks the running release against it. Pricing is pay as you go, with no subscription.\n\n**OpenRouter.** One key for open and proprietary models. Its [ZDR page](https://openrouter.ai/docs/features/zdr) defines ZDR as \"no routine storage of prompts or responses after processing, apart from the CSAM exception,\" and you can require it per request with `\"provider\": {\"zdr\": true}`. [Provider routing](https://openrouter.ai/docs/features/provider-routing) also has `data_collection: \"deny\"` (\"use only providers which do not collect user data\"). The ZDR page notes that self-serve accounts allow \"Safety Retention\" endpoints by default, so check your account settings.\n\n**TrustedRouter.** An OpenAI-compatible router at `api.trustedrouter.com/v1` with a confidential-only host as well ([docs](https://trustedrouter.com/docs)). Its [trust page](https://trustedrouter.com/trust) says \"Public TLS terminates inside the measured GCP Confidential Space workload,\" and you can compare the `image_digest` from `/attestation` with the published one. Per request you can set `\"provider\": {\"min_privacy\": \"zdr\"}` or `\"confidential\"`. The router code is [open source](https://github.com/Lore-Hex/quill-router).\n\n**Phala / RedPill.** RedPill's gateway runs in Intel TDX, and \"the gateway does not store request or response bodies\" ([trust boundary](https://docs.redpill.ai/confidential-ai/trust-boundary.md)). Its E2EE \"is decrypted inside the attested gateway\" ([E2EE docs](https://docs.redpill.ai/confidential-ai/e2e-encryption.md)). GPU-TEE models run on H100, H200 and B300 ([Phala](https://phala.com/confidential-ai)). Proprietary models run on providers that aren't attested, so check `is_tee` per model.\n\n**NEAR AI Cloud.** It runs on \"Intel TDX + NVIDIA GPU TEE\" ([near.ai](https://near.ai)). Its E2EE goes all the way to the model: \"Only model environments holding the attested model key can decrypt your messages\" ([E2EE guide](https://docs.near.ai/cloud/guides/e2ee-chat-completions.md)). OpenAI, Anthropic and Gemini models are offered in an \"Incognito\" mode that runs \"outside a TEE\" ([models](https://docs.near.ai/cloud/models.md)). The verifier is [open source](https://github.com/nearai/nearai-cloud-verifier).\n\n**Tinfoil.** Open-weight models only, at `inference.tinfoil.sh/v1`. \"We do not retain prompt or response content after the response is returned\" ([privacy](https://tinfoil.sh/privacy)). TLS terminates inside the confidential VM, and its [EHBP protocol](https://docs.tinfoil.sh/resources/ehbp.md) \"encrypts HTTP message bodies end-to-end.\" Its SDKs verify attestation for you. Tinfoil [notes](https://docs.tinfoil.sh/sdk/direct-api-access.md) that plain OpenAI-SDK calls skip that check and are \"not recommended for production.\"\n\n**Prem.** Two modes. ZDR mode (`gateway.prem.io/openai`) is contract-based: \"Nobody keeps your prompts, completions, or audio content at any stage of the request\" ([ZDR](https://docs.prem.io/zdr/overview.md)). Confidential mode runs through a local proxy: \"The SDK or local proxy encrypts the request payload before network egress\" ([security model](https://docs.prem.io/security-model.md)), and the Reticle verifier checks the CPU and every GPU.\n\n**Privatemode.** Built by Edgeless Systems. It's \"compatible with the OpenAI and Anthropic APIs\" through a local proxy ([docs](https://docs.privatemode.ai)). \"Prompts are encrypted by the client-side Privatemode proxy and decrypted by a server-side\" proxy on the worker ([encryption](https://docs.privatemode.ai/security/encryption/)). It also says \"the reference values in the manifest can be derived by anyone\" ([attestation](https://docs.privatemode.ai/security/attestation/overview/)).\n\nAll eight speak the Chat Completions format, so the OpenAI SDK you already use keeps working. Here's the Morse version, with a privacy floor on one request:\n\n``` python\nimport os\nfrom openai import OpenAI\n\nclient = OpenAI(\n    base_url=\"https://inference.trymorse.ai/v1\",   # was https://api.openai.com/v1\n    api_key=os.environ[\"MORSE_API_KEY\"],           # was OPENAI_API_KEY\n)\n\nreply = client.chat.completions.create(\n    model=\"z-ai/glm-5.3\",\n    messages=[{\"role\": \"user\", \"content\": \"Summarise this contract clause.\"}],\n    extra_body={\"provider\": {\"min_privacy\": \"tee\"}},  # only attested hardware may serve it\n)\nprint(reply.choices[0].message.content)\n```\n\nThe same pattern works for the others: swap `base_url`, the key and the model id, and move any routing options into `extra_body` (for OpenRouter, `{\"provider\": {\"zdr\": True}}`).\n\n`tee` and `e2ee` tiers, Tinfoil, NEAR AI, Prem (confidential mode), Privatemode and Phala's GPU-TEE models.\nWhatever you pick, run the provider's verifier in CI rather than once by hand. That turns a privacy claim into a check that can fail.\n\n**Is zero data retention the same as a TEE?**\n\nNo. ZDR is a promise not to store your data, usually backed by a contract. A TEE is hardware that stops the operator from reading memory while your request runs, and attestation lets you check which code is running.\n\n**Can I keep using the OpenAI SDK?**\n\nYes, with all eight. You change the base URL, the key and the model id. Prem and Privatemode can also run a local proxy that you point the SDK at.\n\n**What does \"attestation you can check\" mean in practice?**\n\nA command or library fetches the enclave's hardware-signed report and compares its code measurement with a published build. Examples: `npx @askmorse/verify`, `tinfoil attestation verify`, `@phala/aci-verifier`, NEAR AI's verifier and Prem's Reticle.\n\n**Does E2EE mean the provider can't see my prompt?**\n\nIt depends on where the encryption ends. On RedPill it ends at the attested gateway. On Morse's `e2ee` tier, NEAR AI, Tinfoil, Prem's confidential mode and Privatemode, it ends inside the enclave that runs the model.\n\n**Which of these offer GPT and Claude?**\n\nOpenRouter, TrustedRouter, Phala/RedPill and NEAR AI. Morse's API, Tinfoil, Prem and Privatemode serve open-weight models such as DeepSeek, GLM, Kimi, Qwen and gpt-oss.\n\n**How do I make sure a request is never served at a lower privacy level?**\n\nSet the floor on the request. Morse and TrustedRouter have `provider.min_privacy`, OpenRouter has `provider.zdr`, and RedPill has `provider.zdr`, which fails closed with a 404.\n\n**How does Morse's API price requests?**\n\nPay as you go, with no subscription. Each model's price per million tokens is listed on the [models page](https://trymorse.ai/inference/models?src=seo-devto).", "url": "https://wpnews.pro/news/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable", "canonical_source": "https://dev.to/trymorse/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable-enclaves-2iln", "published_at": "2026-11-21 01:25:50+00:00", "updated_at": "2026-11-21 01:42:16.526770+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-infrastructure", "ai-safety", "ai-tools"], "entities": ["Morse", "OpenRouter", "TrustedRouter", "Phala", "RedPill", "NEAR AI Cloud", "Tinfoil", "Prem"], "also_reported_by": [], "takeaway": "Of the seven OpenAI-compatible APIs surveyed, all offer zero retention, but only Morse, TrustedRouter, Phala/RedPill, NEAR AI Cloud, Tinfoil, Prem and Privatemode document TEE-backed enclaves with checkable attestation, and OpenRouter's self-serve accounts permit safety-retention endpoints unless `z", "recommendation": {"advice": "When routing prompts through an OpenAI-compatible API, require zero retention per request and verify the enclave's attestation against published source before trusting a provider's privacy claims.", "playbook": "agent-security", "playbook_url": "https://wpnews.pro/playbooks/agent-security"}, "alternates": {"html": "https://wpnews.pro/news/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable", "markdown": "https://wpnews.pro/news/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable.md", "text": "https://wpnews.pro/news/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable.txt", "jsonld": "https://wpnews.pro/news/private-openai-compatible-llm-apis-compared-zero-retention-tee-and-verifiable.jsonld"}}