Enter opens the highlighted result. Esc closes.

Best open-weight LLM for reasoning

DeepSeek V4.1 Flash 1017 leads
80.4
models ranked
55
downloads of #1, 30 days
42.1K
lead over #2
+2.1

Reasoning Index · read October 21

  1. New DeepSeek V4.1 Flash 1017 October 17
  2. New E Glimpse October 17
  3. New Mercury 3 October 16
  4. New R Beam 1.0 501B A23B October 16

As of October 21, 2026, DeepSeek V4.1 Flash 1017 leads 55 open-weight models on Reasoning Index with 80.4, ahead of MoonshotAI Kimi K3 (78.3) and Z.ai GLM 5.3 (78.2).

The full best LLM for reasoning ranking combines this benchmark with others.

Top 20 of 55, higher is better

Embed this data
Best open-weight LLM for reasoning
Rank Model Reasoning Index Coding evidence Weights Coding index Output $/Mtok
1
DeepSeek V4.1 Flash 1017 DeepSeek
80.4
 deepseek-ai/DeepSeek-V4.1-Flash-1017 70.3 $0.60
2
Kimi K3 Moonshot AI
78.3
 moonshotai/Kimi-K3 62.3 $15.00
3
GLM 5.3 Z.ai
78.2
 zai-org/GLM-5.3 70.6 $4.40
4
DeepSeek V4.1 Flash DeepSeek
75.7
 deepseek-ai/DeepSeek-V4.1-Flash 62.5 $0.78
5
MiMo-V2.6-Pro Xiaomi
74.2
 XiaomiMiMo/MiMo-V2.6-Pro-RL 65.4 $0.87
6
MiniMax M3 MiniMax
73.6
 MiniMaxAI/Minimax-M3 39.2 $1.20
7
Step 5 Preview StepFun
73.4
 stepfun-ai/Step-5-Preview 61.1 $2.70
8
GLM 5.3 Flash Z.ai
72.8
 zai-org/GLM-5.3-Flash 57.7 $0.50
9
Kimi K2.6 Moonshot AI
72.2
 moonshotai/Kimi-K2.6 52.1 $4.00
10
DeepSeek V4 Pro 0813 DeepSeek
71.7
 deepseek-ai/DeepSeek-V4-Pro-0813 56.1 $1.98
11
DeepSeek V4 Flash 0731 DeepSeek
70.2
 deepseek-ai/DeepSeek-V4-Flash-0731 51.8 $0.28
12
Kimi K2.7 Code Moonshot AI
68.9
 moonshotai/Kimi-K2.7-Code 53.4 $4.00
13
Muse Spark Open Meta
67.1
 meta-models/Muse-Spark-Open-816B-A42B 47.4 $4.25
14
GLM 5.1 Z.ai
66.4
 zai-org/GLM-5.1 49.2 $4.40
15
Qwen3.5 397B A17B Qwen
63.9
 Qwen/Qwen3.5-397B-A17B Not enough data $2.34
16
DeepSeek V4 Flash Vision Exp DeepSeek
63.7
 deepseek-ai/DeepSeek-V4-Flash-Vision-Exp Not enough data $1.32
17
T Inkling Thinking Machines
63.1
 thinkingmachines/Inkling 31.5 $4.05
18
Nemotron 3 Ultra NVIDIA
63.0
 nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 30.0 $2.40
19
GLM 5.2 Z.ai
62.2
 zai-org/GLM-5.2 61.1 $4.40
20
GLM 4.7 Z.ai
61.2
 zai-org/GLM-4.7 55.1 $2.20

How this list is built

Generated from the dataset on every refresh. Nobody picks the order by hand.

Models on the Reasoning Index board that OpenRouter links to a public Hugging Face repository, by their Reasoning Index result. We do not check the license; read it on the repository page.

Other lists

Rankings beyond coding · Methodology