FITMYLLM · NOVEMBER 4, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
127.49B
Parameters (5.1B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 256K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 38.5 GB
38.4 + 0.1 KV
low IQ2_M 2.93 47.3 GB
47.2 + 0.1 KV
low Q2_K 3.16 50.9 GB
50.8 + 0.1 KV
low IQ3_XXS 3.25 52.4 GB
52.3 + 0.1 KV
low IQ3_XS 3.5 56.4 GB
56.3 + 0.1 KV
low Q3_K_S 3.64 58.6 GB
58.5 + 0.1 KV
low IQ3_M 3.76 60.5 GB
60.4 + 0.1 KV
low Q3_K_M 4 64.3 GB
64.2 + 0.1 KV
low Q3_K_L 4.3 69.1 GB
69.0 + 0.1 KV
moderate IQ4_XS 4.46 71.7 GB
71.6 + 0.1 KV
moderate Q4_K_S 4.67 75.0 GB
74.9 + 0.1 KV
moderate Q4_K_M 4.89 78.5 GB
78.4 + 0.1 KV
good Q5_K_S 5.57 89.3 GB
89.3 + 0.1 KV
good Q5_K_M 5.7 91.4 GB
91.3 + 0.1 KV
good Q6_K 6.56 105.1 GB
105.0 + 0.1 KV
excellent Q8_0 8.5 136.0 GB
135.9 + 0.1 KV
lossless FP16 16 255.6 GB
255.5 + 0.1 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 LING 3.0 FLASH 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_XXS — 38.4 GB VRAM IQ2_M — 47.2 GB VRAM Q2_K — 50.8 GB VRAM IQ3_XXS — 52.3 GB VRAM IQ3_XS — 56.3 GB VRAM Q3_K_S — 58.5 GB VRAM IQ3_M — 60.4 GB VRAM Q3_K_M — 64.2 GB VRAM Q3_K_L — 69.0 GB VRAM IQ4_XS — 71.6 GB VRAM Q4_K_S — 74.9 GB VRAM Q4_K_M — 78.4 GB VRAM Q5_K_S — 89.3 GB VRAM Q5_K_M — 91.3 GB VRAM Q6_K — 105.0 GB VRAM Q8_0 — 135.9 GB VRAM FP16 — 255.5 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:127b-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 Ling 3.0 Flash
Build Hardware for Ling 3.0 Flash ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Ling 3.0 Flash — 127.49B MoE. ▸ SPECIFICATIONS
PARAMETERS 127.49B (5.1B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 256K tokens
CAPABILITIES chat, coding, reasoning, multilingual, math, agentic, tool_use
RELEASE DATE 2026-08-04
PROVIDER Ant Group
FAMILY other ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 38.4 GB 65% IQ2_M 2.93 47.2 GB 75% Q2_K 3.16 50.8 GB 78% IQ3_XXS 3.25 52.3 GB 82% IQ3_XS 3.5 56.3 GB 84% Q3_K_S 3.64 58.5 GB 85% IQ3_M 3.76 60.4 GB 86% Q3_K_M 4 64.2 GB 88% Q3_K_L 4.3 69.0 GB 90% IQ4_XS 4.46 71.6 GB 92% Q4_K_S 4.67 74.9 GB 93% Q4_K_M 4.89 78.4 GB 94% Q5_K_S 5.57 89.3 GB 96% Q5_K_M 5.7 91.3 GB 96% Q6_K 6.56 105.0 GB 97% Q8_0 8.5 135.9 GB 100% FP16 16 255.5 GB 100%
§ 01 BENCHMARK SCORES
GPQA Diamond 85.5
LiveCodeBench 82.8
AIME 93.2
HLE 23.7
aa_scicode 42.0
aa_lcr 73.0
aa_terminal_bench_v2 55.4
aa_terminal_bench_v4 0.0
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback