NEAR AI
Standalone network · Tier 2 · API Live
Last checked 2026-10-12Record status: Unverified@near_aiJSON ↗ Seeded record — this provider has not yet been through the independent verification pass. Treat every claimed/unverified field below as unconfirmed.
Identity
| Category | Standalone network |
| Tier | 2 |
| Record status | Unverified |
| First tracked | 2026-08-24 |
| Last checked | 2026-10-12 |
| Last changed | 2026-10-12 |
| Website | https://near.ai |
| Aliases | near_ai, nearprotocol, IronClaw |
NEAR AI is NEAR Protocol's user-owned AI infrastructure effort, most recently
visible through IronClaw, a WASM-sandboxed agent runtime, and a confidential GPU
marketplace that leverages TEEs so compute providers cannot read the workloads
they run. NEAR token stake is claimed to convert into compute credits.
The roster describes these as emerging/live features rather than a mature,
long-running product; no direct probe or attestation check has been run for this
seed batch.
API
Privacy
Scale
On-chain
Change log
- 2026-08-24 — record: seeded from provider roster v1.0 (Unverified) (source)
- 2026-09-14 — api.models: X claim (a_launch): The NEAR AI Cloud API is OpenAI-compatible. Point your client at a new base URL and the rest of your code stays as is. (Claimed) (source)
- 2026-09-14 — scale.tokens_per_day: X claim (c_scale): Abound serves 800K+ users and has moved $300M+ in remittances. Its AI Financial Autopilot runs on NEAR AI (Claimed) (source)
- 2026-09-21 — api.models: X claim (a_launch): NEAR AI Cloud's confidential inference is now live on @say_gm_'s confidential tier. SayGm reaches dozens of models through one API key, and runs its own routing inside an Intel TDX enclave rather than on ordinary servers. (Claimed) (source)
- 2026-09-21 — privacy.mechanism: X claim (e_privacy): SayGm protects the route. NEAR AI protects the environment running the model: Intel TDX virtual machines paired with NVIDIA confidential-computing GPUs. No account or API key needed to request our attestation report and check it against Intel and NVIDIA. (Claimed) (source)
- 2026-10-05 — api.models: X claim (a_launch): Wanted to move away from OpenAI and Anthropic, so I built a #VSCode extension that brings @near_ai private inference directly into #Copilot Use SOTA models and keep your chats private and secure, with cryptographic attestations to verify that Available @ https://marketplace.visualstudio.com/items?itemName=NEARDev.near-ai-for-copilot (Claimed) (source)
- 2026-10-05 — scale.tokens_per_day: X claim (c_scale): NEAR AI has claimed first place on the Lean Eval v1 leaderboard. Lean Eval v1 is an active benchmark maintained by Lean FRO that contains extremely hard formalization problems. We reached first place despite joining the competition just four weeks ago, a full month later than other top participants. ... The system that enabled us to claim first place is almost entirely powered by DeepSeek V4.1 Flash, and thus is (a) extremely cheap, and (b) has no dependencies on third parties because the model is open weight. (Claimed) (source)
- 2026-10-12 — privacy.mechanism: X claim (e_privacy): @near_ai's Confidential TEE tier runs models on protected CPUs and GPUs, isolating inference data from host operators. NEAR AI clients can verify gateway and model attestations. Optional end-to-end encryption also hides prompt content from the gateway. (Claimed) (source)
The change log is append-only: corrections are recorded as new entries, never rewrites. Full feed via /api/v1/changes.
Sources
- [roster] docs/decentralized-inference-provider-roster.md (2026-08-24)
- [website] https://near.ai (2026-08-24)
- [x-account] https://x.com/near_ai (2026-08-24)