Pokee-Isaac 28B Launches: 10M Token Context on a Single GPU at $0.15/M
Kyrannio · x · 2026-08-05
Pokee AI released the Pokee-Isaac 28B model, claiming to be the world's first real frontier-class agentic model supporting a 10 million token context. It uses a non-decoder-only architecture and can be deployed on a single GPU (starting from RTX 4090).
Core Performance & Metrics:
- Achieves 93.3% on RULER at 10M tokens.
- Up to 137K tokens/s prefill on a single B200.
- Leads in BFCL v4 and τ³-bench evaluations.
- Reports the lowest combined attack success rate (35.6% ASR on DTAP) among evaluated models.
Pricing & Deployment:
- Priced at $0.15/M input tokens and $1/M output tokens.
- Supports VPC, on-premises, and on-device deployment with Day-0 support for vLLM and SGLang.
Related event: Pokee-Isaac 28B Launches: 10M Context on Single GPU, Tops BFCL v4(5 posts)→
More from Infra
- Nvidia's Strategic Shift: From Chip Sales to Cloud Revenue Sharing — GavinSBaker · 2026-08-05
- llama.cpp PR Caches Hot MoE Experts on GPU, Doubling Inference Speed on 8GB VRAM — BTA_Labs · 2026-08-05
- Big Tech Locks in Over $1 Trillion in Future AI Data Center Lease Payments — Polymarket · 2026-08-05
- YC-Backed Lamb Labs Claims 63x Higher Efficiency for AI Inference Chips vs GPUs — ycombinator · 2026-08-05
- DSpark Open-Sources Speculative Decoding Path for Kimi K3 — ying11231 · 2026-08-05
- US Drafts Ban on Chinese Datacenter Components as Europe Pushes for Tech Sovereignty — nordicinst · 2026-08-05