Is Bonsai 2 27B REALLY 98% of Qwen 3.8 27B fp16? Local AI Test

Description: We evaluate inference speed (~68–70 tokens/sec) of Ternary Bonsai 2 27B and put its agentic coding capabilities to the test with Hermes Agent to generate a full arcade game suite—comparing the results head-to-head against full fp16 Qwen 3.8 27B.

🔧 Guides & Links:
• GPU Pricing & Specs Tracker: https://digitalspaceport.com/local-ai-gpu-price-and-performance/
• Test the generated arcade games: https://arcade.digitalspaceport.com
• Proxmox LXC Local AI Guide: https://digitalspaceport.com/how-to-setup-an-ai-server-homelab-beginners-guides-ollama-and-openwebui-on-proxmox-lxc/

Threadripper WRX80 Build
MOBO - WRX80
Asrock Creator https://geni.us/ASRock_wrx80_creator
Gigabyte MC62-G40 https://geni.us/MC62-G40_MOBO

CPU - AMD Threadripper PRO
3945WX https://geni.us/AMD-TR-3945WX

GPUs
3090 24GB https://geni.us/GPU3090
3060 12GB https://geni.us/3060_GPU_12GB
4090 24GB https://geni.us/4090_24GB_GPU

RAM
256GB DDR4 DIMMs https://geni.us/256GB_DDR4_RAM

RISER CABLES
x16 wide PCIe Risers https://geni.us/PCIe4_Riser_Cable

Rack Frame Chassis https://geni.us/GPU_Rack_Frame

PSUs
1500W Power Supply https://geni.us/Corsair_HX1500iPSU
1000W Power Supply https://geni.us/1000W_PSU

H170i CAPELLIX Liquid CPU Cooler https://geni.us/iCUE_H170i_Capellix

The Only Local AI GPUs Worth Buying Right Now Video https://youtu.be/Oa3MUGsFays
Local Ai Server Build Playlist https://www.youtube.com/playlist?list=PLarJAzZsWRGDktdCjcN51G4xVkvvjC0ZB

Ways to Support:
🚀 Join as a member for members-only content and extra perks https://www.youtube.com/c/digitalspaceport/join
☕ Buy Me a Coffee https://www.buymeacoffee.com/digitalspaceport
🔳 Patreon https://www.patreon.com/digitalspaceport
👍 Subscribe youtube.com/c/digitalspaceport?sub_confirmation=1
🌐 Check out the Website https://digitalspaceport.com

TIMESTAMPS:
00:00 - Introduction: Ternary Bonsai 2 27B Overview & Claims
00:48 - Setup & Installation in Proxmox LXC Container
01:58 - GPU Market Price Check (RTX 3090 / 3060)
02:47 - llama-server Config & Environment Variables
03:55 - VRAM Usage & Memory Footprint on RTX 3090
04:24 - Speed & Generation Test: Animated SVG Cat (~69 tok/s)
07:44 - Hermes Agentic Coding Setup & Prompt
09:10 - Testing the Generated Arcade Games
12:33 - Comparing Quality vs Qwen 3.8 27B FP16
13:51 - Final Verdict: Speed vs Quality Trade-offs


*****
As an Amazon Associate I earn from qualifying purchases.
When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
***** Close description