Kev 4B Decision Model Lands on Redpill: TEE Private Inference at $0.042/M Input
bgmshana · x · 2026-10-14
The author announces that Kev 4B — a fast, cheap open-weight decision model close to Jev, fine-tuned from Qwen3.5 4B Base — is now served via dstack on Phala Network and Redpill in TEE enclaves. It skips chat entirely and returns structured decisions (yes/no probability, choices, scale scores) at $0.042/M input tokens (output free) with an 8,192-token context. The Redpill gateway is OpenAI SDK-compatible, supports Private AI Proxy for Claude Code, Codex and other agents, and provides receipt verification with attestation reports.
Related event: Open-Source Decision Model Kev-4B Launches with TEE Private Inference(3 posts)→
More from Infra
- RISC-V crypto accelerator SoC gets full place-and-route done with ~$2,000 of LLM tokens — IgorCarron · 2026-10-14
- SK Hynix has struggled with HBM4 back-end issues for over a year, Samsung tipped to lead HBM market next year — zephyr_z9 · 2026-10-14
- Startup Builds $18,000 Local AI Server With 128GB VRAM, Runs 16 Concurrent Sessions — bendee983 · 2026-10-14
- A used dual-RTX-3090 local LLM rig that isn't much faster — why? — jonejy · 2026-10-14
- Photonic Training of a 1B GPT-2 Without Backprop Lands in PNAS — IgorCarron · 2026-10-14
- Xinjiang's 167 GW green power, 16ms to Chongqing: the grid edge in AI race — 96Stats · 2026-10-14