Skip to content
Local AI Wizard
Your machine
Theme

Local text models

Code · Chat, writing, code, translation, classification and extraction.

Quality · Coding

LiveBench 0–100 · higher is better

  • Color = publisher
  • Memory under review

Source: LiveBench · release of Jun 25, 2026 · checked Oct 18, 2026 · 37 open models without a score in this release are not drawn (no score ≠ 0)

Top score

Smaug Agentic: 82.5 on LiveBench Coding

Lightest scored

Qwen3.8 27B: ~21 GB Q4, model only, 75.7

Licenses

13 of 23 scored open models use MIT or Apache 2.0

Catalog

37 more repositories without a LiveBench score (no score ≠ 0)

Choose your machine and the charts and tables mark what fits in its memory.

23 scored models · LiveBench Jun 25, 2026 

Text models
Smaug AgenticAbacus AI · Publisher terms: Kimi K382.52,800~2.1 TBPublisher terms: Kimi K3
Kimi K3Moonshot AI · Publisher terms: Kimi K381.42,800~2.1 TBPublisher terms: Kimi K3
DeepSeek V4.1 Flash · maxDeepSeek · MIT80.0Not publishedMemory estimate under reviewMIT
GLM-5.2Z.ai · MIT79.7744~558 GBMIT
GLM-5.3Z.ai · Publisher terms: GLM-5.379.0744~558 GBPublisher terms: GLM-5.3
GLM-5.3 FlashZ.ai · MIT79.0320~240 GBMIT
Kimi K2.6 · thinkingMoonshot AI · Publisher terms: Modified MIT78.61,000~750 GBPublisher terms: Modified MIT
DeepSeek V4 Pro · 0813DeepSeek · MIT77.21,600~1.2 TBMIT
Smaug FlashAbacus AI · MIT77.1Not publishedMemory estimate under reviewMIT
Qwen3.8 27BAlibaba · Apache 2.075.727~21 GBApache 2.0
Smaug MiniAbacus AI · Apache 2.075.427~21 GBApache 2.0
DeepSeek V4 Flash · 0731DeepSeek · MIT75.0304~228 GBMIT
Kimi K2.7 CodeMoonshot AI · Publisher terms: Modified MIT74.01,000~750 GBPublisher terms: Modified MIT
Qwen3.8 MaxAlibaba · Publisher terms: Qwen3.8 Max72.92,400~1.8 TBPublisher terms: Qwen3.8 Max
Qwen3.8 Flash NextAlibaba · Publisher terms: Qwen Community 1.072.6Not publishedMemory estimate under reviewPublisher terms: Qwen Community 1.0
Qwen3.6 27BAlibaba · Apache 2.071.827~21 GBApache 2.0
Inkling · xhighThinking Machines · Apache 2.071.0975~732 GBApache 2.0
Nemotron 3 Ultra 550B A55BNVIDIA · OpenMDW 1.170.7550~413 GBOpenMDW 1.1
DeepSeek V4 ProDeepSeek · MIT70.01,600~1.2 TBMIT
DeepSeek V4 FlashDeepSeek · MIT69.2284~213 GBMIT
DeepSeek V4 Flash Vision · expDeepSeek · MIT68.2Not publishedMemory estimate under reviewMIT
MiniMax M3MiniMax · Publisher terms: MiniMax Community68.2428~321 GBPublisher terms: MiniMax Community
Nemotron 3.5 Lightning 30B A3BNVIDIA · OpenMDW 1.145.830~23 GBOpenMDW 1.1