FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K
Quant Bits VRAM @ 16K Quality IQ2_M 2.93 32.8 GB
29.1 + 3.8 KV
low Q2_K 3.16 35.0 GB
31.3 + 3.8 KV
low IQ3_XXS 3.25 35.9 GB
32.2 + 3.8 KV
low IQ3_XS 3.5 38.4 GB
34.6 + 3.8 KV
low Q3_K_S 3.64 39.7 GB
36.0 + 3.8 KV
low IQ3_M 3.76 40.9 GB
37.1 + 3.8 KV
low Q3_K_M 4 43.2 GB
39.5 + 3.8 KV
low Q3_K_L 4.3 46.2 GB
42.4 + 3.8 KV
moderate IQ4_XS 4.46 47.7 GB
44.0 + 3.8 KV
moderate Q4_K_S 4.67 49.8 GB
46.0 + 3.8 KV
moderate Q4_K_M 4.89 51.9 GB
48.2 + 3.8 KV
good Q5_K_S 5.57 58.5 GB
54.8 + 3.8 KV
good Q5_K_M 5.7 59.8 GB
56.1 + 3.8 KV
good Q6_K 6.56 68.2 GB
64.4 + 3.8 KV
excellent Q8_0 8.5 87.1 GB
83.4 + 3.8 KV
lossless FP16 16 160.2 GB
156.5 + 3.8 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 INTERNVL3 78B 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 IQ2_M — 29.1 GB VRAM Q2_K — 31.3 GB VRAM IQ3_XXS — 32.2 GB VRAM IQ3_XS — 34.6 GB VRAM Q3_K_S — 36.0 GB VRAM IQ3_M — 37.1 GB VRAM Q3_K_M — 39.5 GB VRAM Q3_K_L — 42.4 GB VRAM IQ4_XS — 44.0 GB VRAM Q4_K_S — 46.0 GB VRAM Q4_K_M — 48.2 GB VRAM Q5_K_S — 54.8 GB VRAM Q5_K_M — 56.1 GB VRAM Q6_K — 64.4 GB VRAM Q8_0 — 83.4 GB VRAM FP16 — 156.5 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:78b-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 InternVL3 78B
Build Hardware for InternVL3 78B ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
InternVL3 78B — 78B Dense. ▸ SPECIFICATIONS
PARAMETERS 78B
ARCHITECTURE Dense Transformer
CONTEXT LENGTH 32K tokens
CAPABILITIES chat, vision, reasoning
RELEASE DATE 2025-04-15
PROVIDER Shanghai AI Lab
FAMILY other ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_M 2.93 29.1 GB 75% Q2_K 3.16 31.3 GB 78% IQ3_XXS 3.25 32.2 GB 82% IQ3_XS 3.5 34.6 GB 84% Q3_K_S 3.64 36.0 GB 85% IQ3_M 3.76 37.1 GB 86% Q3_K_M 4 39.5 GB 88% Q3_K_L 4.3 42.4 GB 90% IQ4_XS 4.46 44.0 GB 92% Q4_K_S 4.67 46.0 GB 93% Q4_K_M 4.89 48.2 GB 94% Q5_K_S 5.57 54.8 GB 96% Q5_K_M 5.7 56.1 GB 96% Q6_K 6.56 64.4 GB 97% Q8_0 8.5 83.4 GB 100% FP16 16 156.5 GB 100%
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 