Weekly archive

New local AI models, Jan 20–26, 2025

15 open models released Jan 20–26, 2025, with how each runs on five reference devices.

DeepSeek R1 (671B)

DeepSeek · LLM · 671B (37.0B active) · released Jan 20

77Great quality (A) · 77 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
Memory, every quant and how to run it Check your own device
Selene 1 Mini 8B

Atla · Decision · 8.0B · released Jan 22

79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Llama 70B

DeepSeek · LLM · 71B · released Jan 20

68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Qwen 32B

DeepSeek · LLM · 33B · released Jan 20

68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Won't runover the app memory limit
- Laptop MacBook Pro M5 Pro 64 GB Runs well 12.8 tok/s
- Gaming PC RTX 5090 32 GB Runs welltight fit 46.5 tok/s
- Edge board Jetson Orin NX 16 GB Won't run
- Workstation Mac Studio 512 GB Runs well 26.4 tok/s
Memory, every quant and how to run it Check your own device
GTE Reranker ModernBERT Base

Alibaba · Decision · 150M · released Jan 20

70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Qwen 14B

DeepSeek · LLM · 15B · released Jan 20

61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Won't runover the app memory limit
- Laptop MacBook Pro M5 Pro 64 GB Runs well 22.0 tok/s
- Gaming PC RTX 5090 32 GB Runs great 75.8 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 11.0 tok/s
- Workstation Mac Studio 512 GB Runs great 40.7 tok/s
Memory, every quant and how to run it Check your own device
Lumina-Image 2.0

Shanghai AI Lab · Image · 2.6B · released Jan 22

62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Very slowencoder unloaded after use 2.8 min
- Laptop MacBook Pro M5 Pro 64 GB Slow 68 s
- Gaming PC RTX 5090 32 GB Runs great 5.9 s
- Edge board Jetson Orin NX 16 GB Slow 63 s
- Workstation Mac Studio 512 GB Runs well 30 s
Memory, every quant and how to run it Check your own device
Qwen2.5 14B Instruct 1M

Alibaba · LLM · 15B · released Jan 23

59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Won't runover the app memory limit
- Laptop MacBook Pro M5 Pro 64 GB Runs well 22.0 tok/s
- Gaming PC RTX 5090 32 GB Runs great 75.8 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 11.0 tok/s
- Workstation Mac Studio 512 GB Runs great 40.7 tok/s
Memory, every quant and how to run it Check your own device
Qwen2.5-VL 7B Instruct

Alibaba · LLM · 8.3B · released Jan 26

52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs welltight fit 11.7 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 39.1 tok/s
- Gaming PC RTX 5090 32 GB Runs great 62.3 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 15.2 tok/s
- Workstation Mac Studio 512 GB Runs great 53.2 tok/s
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Llama 8B

DeepSeek · LLM · 8.0B · released Jan 20

50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs welltight fit 12.1 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 40.4 tok/s
- Gaming PC RTX 5090 32 GB Runs great 64.3 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 15.7 tok/s
- Workstation Mac Studio 512 GB Runs great 54.9 tok/s
Memory, every quant and how to run it Check your own device
Qwen2.5 7B Instruct 1M

Alibaba · LLM · 7.6B · released Jan 23

50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs well 12.7 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 42.6 tok/s
- Gaming PC RTX 5090 32 GB Runs great 67.7 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 16.5 tok/s
- Workstation Mac Studio 512 GB Runs great 57.8 tok/s
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Qwen 7B

DeepSeek · LLM · 7.6B · released Jan 20

46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs well 12.7 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 42.6 tok/s
- Gaming PC RTX 5090 32 GB Runs great 67.7 tok/s
- Edge board Jetson Orin NX 16 GB Runs well 16.5 tok/s
- Workstation Mac Studio 512 GB Runs great 57.8 tok/s
Memory, every quant and how to run it Check your own device
Qwen2.5-VL 3B Instruct

Alibaba · LLM · 3.8B · released Jan 26

34Basic quality (D) · 34 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs well 25.8 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 44.1 tok/s
- Gaming PC RTX 5090 32 GB Runs great 138 tok/s
- Edge board Jetson Orin NX 16 GB Runs great 33.5 tok/s
- Workstation Mac Studio 512 GB Runs great 54.2 tok/s
Memory, every quant and how to run it Check your own device
DeepSeek R1 Distill Qwen 1.5B

DeepSeek · LLM · 1.8B · released Jan 20

22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs great 45.0 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 42.8 tok/s
- Gaming PC RTX 5090 32 GB Runs great 290 tok/s
- Edge board Jetson Orin NX 16 GB Runs great 49.0 tok/s
- Workstation Mac Studio 512 GB Runs great 114 tok/s
Memory, every quant and how to run it Check your own device
SmolVLM 500M Instruct

Hugging Face · LLM · 0.5B · released Jan 20

10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking.

Runs on
- Phone iPhone 17 Pro Runs great 44.6 tok/s
- Laptop MacBook Pro M5 Pro 64 GB Runs great 149 tok/s
- Gaming PC RTX 5090 32 GB Runs great 1,012 tok/s
- Edge board Jetson Orin NX 16 GB Runs great 57.8 tok/s
- Workstation Mac Studio 512 GB Runs great 398 tok/s
Memory, every quant and how to run it Check your own device

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.