FITMYLLM · JULY 28, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
Quantization Options Select your GPU for speed estimates Quant Bits VRAM @ 8K Quality IQ3_XXS 3.25 15.4 GB
14.5 + 0.9 KV
low IQ3_XS 3.5 16.5 GB
15.5 + 0.9 KV
low Q3_K_S 3.64 17.1 GB
16.1 + 0.9 KV
low IQ3_M 3.76 17.6 GB
16.7 + 0.9 KV
low Q3_K_M 4 18.6 GB
17.7 + 0.9 KV
low Q3_K_L 4.3 19.9 GB
19.0 + 0.9 KV
moderate IQ4_XS 4.46 20.6 GB
19.7 + 0.9 KV
moderate Q4_K_S 4.67 21.5 GB
20.6 + 0.9 KV
moderate Q4_K_M 4.89 22.5 GB
21.5 + 0.9 KV
good Q5_K_S 5.57 25.4 GB
24.4 + 0.9 KV
good Q5_K_M 5.7 25.9 GB
25.0 + 0.9 KV
good Q6_K 6.56 29.6 GB
28.7 + 0.9 KV
excellent Q8_0 8.5 38.0 GB
37.0 + 0.9 KV
lossless FP16 16 70.2 GB
69.3 + 0.9 KV
lossless
Select your GPU above to see speed estimates and compatibility for each quantization.
Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise → ▸ READY TO RUN THIS? RENT BY THE HOUR
RENT A GPU AND RUN NOUS CAPYBARA 34B NOW
Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.
Community Ratings Chat Coding Reasoning Creative Vision Roleplay Agentic
Loading ratings...
Run this model IQ3_XXS — 14.5 GB VRAM IQ3_XS — 15.5 GB VRAM Q3_K_S — 16.1 GB VRAM IQ3_M — 16.7 GB VRAM Q3_K_M — 17.7 GB VRAM Q3_K_L — 19.0 GB VRAM IQ4_XS — 19.7 GB VRAM Q4_K_S — 20.6 GB VRAM Q4_K_M — 21.5 GB VRAM Q5_K_S — 24.4 GB VRAM Q5_K_M — 25.0 GB VRAM Q6_K — 28.7 GB VRAM Q8_0 — 37.0 GB VRAM FP16 — 69.3 GB VRAM
Ollama llama.cpp vLLM LM Studio KoboldCpp Jan Docker
▸ Easiest way to get started · Beginners
DOCS ↗ curl -fsSL https://ollama.com/install.sh | shCOPY
$ ollama run other:34b-q4_K_MCOPY
Tag may need adjustment — check ollama.com/library/other for available tags.
▸ SETUP GUIDE >_
Auto-setup with fitmyllm CLI Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.
Auto-detect GPU Live tok/s in chat Speed benchmarks 9 inference engines
GPUs that can run this model At Q4_K_M quantization. Sorted by minimum VRAM.
Find the best GPU for Nous Capybara 34B
Build Hardware for Nous Capybara 34B Nous Capybara 34B — long-context model designed for multi-turn conversations.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Nous Capybara 34B — 34.4B Dense. ▸ SPECIFICATIONS
PARAMETERS 34.4B
ARCHITECTURE Dense Transformer
CONTEXT LENGTH 8K tokens
CAPABILITIES chat
RELEASE DATE 2023-12-01
PROVIDER Nous Research
FAMILY other ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ3_XXS 3.25 14.5 GB 82% IQ3_XS 3.5 15.5 GB 84% Q3_K_S 3.64 16.1 GB 85% IQ3_M 3.76 16.7 GB 86% Q3_K_M 4 17.7 GB 88% Q3_K_L 4.3 19.0 GB 90% IQ4_XS 4.46 19.7 GB 92% Q4_K_S 4.67 20.6 GB 93% Q4_K_M 4.89 21.5 GB 94% Q5_K_S 5.57 24.4 GB 96% Q5_K_M 5.7 25.0 GB 96% Q6_K 6.56 28.7 GB 97% Q8_0 8.5 37.0 GB 100% FP16 16 69.3 GB 100%
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback