open god mode ▮

Open-weights LLM ranking — overall

# new releases

Ling 3.0 Flash Fin inclusionai · Sep 2026 127.5B · ↑ trending K2 Horizon 375B A23B IFM · Sep 2026 379.2B · ↑ trending Hy4 preview tencent · Aug 2026 780B · ↑ trending Qwen3.8 Flash Next qwen · Aug 2026 180B · ↑ trending

scored as 70% aa_intelligence · 30% bench_composite (renormalized when signals are missing) — methodology

# model score $/M out org params ctx prov
1 kimi-k3 ● 100.0 $12.75 moonshotai 2800B 1024k 24
2 deepseek-v4-pro ● 96.1 $0.87 deepseek-ai 862B 1024k 25
3 glm-5.2 ● 96.1 $1.08 zai-org 753B 1024k 43
4 kimi-k2.6 ● 92.5 $1.20 moonshotai 1059B (A32B) 256k 28
5 hy3 ● 92.1 $0.58 tencent 298.8B 256k 5
6 minimax-m3 ● 91.7 $0.96 minimax 427B 1024k 18
7 kimi-k2.7-code ● 90.0 $3.00 moonshotai 1059B (A32B) 256k 24
8 deepseek-v4-flash ● 86.9 $0.17 deepseek-ai 158B 1024k 25
9 glm-5 ● 86.3 $1.92 zai-org 754B 200k 14
10 glm-5.1 ● 85.2 $2.86 zai-org 754B 200k 22
11 minimax-m2.5 ● 81.4 $0.95 minimax 229B (A10B) 200k 17
12 kimi-k2.5 ● 80.8 $2.25 moonshotai 1059B (A32B) 256k 14
13 qwen3.6-27b ● 79.9 $0.50 qwen 27.8B 256k 13
14 qwen3.5-27b ● 74.5 $1.56 qwen 27.8B 256k 7
15 glm-4.7 ● 74.1 $1.75 zai-org 358B 200k 11
16 qwen3.6-35b-a3b ● 73.1 $0.50 qwen 36B (A3.3B) 256k 16
17 qwen3.5-397b-a17b ● 73.0 $2.00 qwen 403B (A17B) 256k 18
18 qwen3.5-122b-a10b ● 70.4 $1.75 qwen 125B (A10B) 256k 6
19 deepseek-v3.2 ● 68.6 $0.31 deepseek-ai 671B (A37B) 160k 25
20 kimi-k2-thinking ● 66.8 $1.20 moonshotai 1000B (A32B) 256k 7
21 minimax-m2.1 ● 65.8 $1.20 minimax 229B (A10B) 200k 6
22 kimi-k2-0905 ● 65.5 $2.00 moonshotai 1000B (A32B) 256k 5
23 qwen3.5-35b-a3b ● 64.4 $0.75 qwen 36B (A3B) 256k 9
24 glm-4.6 ● 63.0 $1.75 zai-org 355B (A32B) 200k 8
25 gemma-4-31b-it ● 62.3 $0.34 google 32.7B 256k 22
26 granite-4.2-30b ● 60.0 $0.65 ibm-granite 29.3B — 1
27 gemma-4-26b-a4b-it ● 58.2 $0.22 google 26.5B (A4B) 256k 12
28 glm-4.7-flash ● 57.5 $0.40 zai-org 31.2B 198k 7
29 qwen3-vl-235b-a22b-thinking ● 57.3 $2.50 qwen 235.7B (A22B) 128k 7
30 minimax-m2 ● 56.7 $0.60 minimax 229B (A10B) 200k 5
31 gemma-4-12b-it ● 53.7 — google 12B — 1
32 qwen3-coder-480b-a35b ● 51.6 $1.00 qwen 480B (A35B) 256k 11
33 deepseek-r1 ● 47.0 $2.00 deepseek-ai 671B (A37B) 64k 12
34 glm-4.5 ● 46.8 $2.20 zai-org 355B (A32B) 128k 3
35 gpt-oss-120b ● 45.7 $0.17 openai 117B (A5.1B) 128k 36
36 kimi-k2-instruct ● 44.3 $2.00 moonshotai 1000B (A32B) 128k 6
37 deepseek-v3.1 ● 43.4 $0.95 deepseek-ai 671B (A37B) 160k 15
38 granite-4.2-3b ● 41.8 $0.12 ibm-granite 3.7B — 1
39 qwen3-235b-a22b-instruct-2507 ● 37.8 $0.35 qwen 235.1B (A22B) 256k 16
40 qwen3-vl-30b-a3b-thinking ● 35.9 $1.00 qwen 30.5B (A3B) 256k 3
41 glm-4.5-air ● 35.8 $0.85 zai-org 106B (A12B) 128k 4
42 deepseek-v3 ● 33.8 $0.20 deepseek-ai 671B (A37B) 160k 11
43 qwen3-coder-30b-a3b ● 32.6 $0.26 qwen 30.5B (A3.3B) 256k 7
44 minimax-m1 ● 31.8 $2.20 minimax 456B (A45.9B) 1000k 2
45 qwen3-vl-235b-a22b-instruct ● 31.6 $0.88 qwen 235.7B (A22B) 256k 9
46 qwen3-vl-8b-thinking ● 28.4 $2.10 qwen 8.8B 128k 1
47 llama-4-maverick ● 27.6 $0.49 meta-llama 400B (A17B) 1024k 7
48 qwq-32b ● 25.7 $0.20 qwen 32.8B — 6
49 qwen3-vl-32b-instruct ● 19.3 $0.42 qwen 33.4B 128k 4
50 magistral-small-2506 ● 17.3 $1.50 mistralai 24B — 1
51 qwen3-vl-30b-a3b-instruct ● 16.6 $0.52 qwen 30.5B (A3B) 256k 7
52 nvidia-nemotron-nano-9b-v2 ● 15.8 $0.16 nvidia 8.9B — 2
53 devstral-small-2507 ● 15.7 $0.30 mistralai 24B — 1
54 gemma-3-27b-it ● 10.8 $0.16 google 27B 128k 8
55 qwen3-vl-8b-instruct ● 6.5 $0.46 qwen 8.8B 256k 4
56 qwen3-vl-4b-instruct ● 2.3 — qwen 4.4B — —
— qwen3.8-2.4t-a95b ● 99.0 $6.00 qwen 2446.2B 1024k 8
— qwen3.8-flash-next ● 97.9 — qwen 180B — —
— deepseek-v4-flash-0731 ● 94.8 $0.10 deepseek-ai 158B 1280k 37
— nemotron-3-nano-30b-a3b ● 94.3 $0.20 nvidia 31.6B 256k 6
— motif-3 ● 93.8 — Motif-Technologies 314.8B — —
— k2-horizon-375b-a23b ● 92.8 — IFM 379.2B — —
— nex-n2-pro ● 86.6 $1.00 nex-agi 396.8B 256k 3
— mimo-v2.5-pro ● 85.6 $0.61 xiaomi 1023B 1050k 9
— inkling ● 83.5 $4.05 thinkingmachines 975B 1024k 6
— minimax-m2.7 ● 81.4 $0.96 minimax 229B (A10B) 200k 13
— muse-glimmer-30b ● 80.0 $1.10 meta-models 29.8B 128k 5
— solar-open2-250b ● 79.9 — upstage 250.3B — —
— a.x-k2 ● 75.8 — skt 691.7B — —
— longcat-2.0 ● 70.1 $2.95 meituan-longcat 1776B 1048k 3
— qwen3-next-80b-a3b-thinking ● 69.3 $1.20 qwen 81.3B (A3B) 256k 9
— g9v3-39a5b ● 69.1 — ai9stars 39B — —
— ring-2.6-1t ● 64.9 $2.50 inclusionai 1025.7B — 1
— deepseek-v3.1-terminus ● 63.9 $0.95 deepseek-ai 671B (A37B) 160k 6
— ling-2.6-1t ● 58.8 $2.50 inclusionai 1025.7B — 1
— qwen3-235b-a22b ● 57.1 $0.60 qwen 235B (A22B) 128k 7
— deepseek-r1-0528 ● 51.0 $0.60 deepseek-ai 671B (A37B) 160k 9
— qwen3.5-9b ● 51.0 $0.15 qwen 9.7B 256k 7
— minicpm-sala ● 49.0 — openbmb 9.5B — —
— qwen3-coder-next ● 48.5 $0.80 qwen 79.7B 256k 7
— deepswe-preview ● 44.2 — agentica-org 32.8B — —
— trinity-large-thinking ● 43.3 $0.80 arcee-ai 398.6B 256k 1
— phi-4-reasoning-plus ● 42.8 $0.50 microsoft 14.7B — 1
— nemotron-cascade-2-30b-a3b ● 41.2 $0.60 nvidia 31.6B — 1
— k2-think-v2 ● 40.2 — IFM 72.6B — —
— devstral-small-2512 ● 40.0 — mistralai 24B — —
— glm-4.6v ● 38.7 $0.90 zai-org 107.7B 128k 2
— g9v3-3b ● 36.1 — ai9stars 3B — —
— devstral-2-2512 ● 35.9 $2.00 mistralai 123B 256k 2
— mistral-large-3 ● 35.1 $0.75 mistralai 675B (A41B) 256k 5
— solar-open-100b ● 33.5 — upstage 102.7B — —
— qwen3-32b ● 32.6 $0.25 qwen 32.8B 128k 12
— gpt-oss-20b ● 32.0 $0.10 openai 21B (A3.6B) 128k 23
— qwen2.5-coder-32b ● 28.4 $0.16 qwen 32.8B 32k 5
— k2-v2 ● 27.8 — IFM — — —
— ling-2.6-flash ● 26.8 $0.30 inclusionai 107.5B — 1
— qwen3-30b-a3b ● 25.6 $0.30 qwen 30.5B (A3.3B) 128k 3
— ling-1t ● 22.7 — inclusionai 1000B (A50B) — —
— minicpm5-1b ● 21.6 — openbmb 1.1B — —
— mistral-small-3.2-24b ● 19.1 $0.20 mistralai 24B 128k 4
— llama-4-scout ● 16.5 $0.29 meta-llama 109B (A17B) 1280k 7
— qwen2.5-72b-instruct ● 14.4 $0.30 qwen 72.7B 32k 6
— llama-3.3-70b-instruct ● 13.4 $0.20 meta-llama 70.6B 128k 22
— mistral-large-2411 ● 11.3 $6.00 mistralai 123B — 2
— granite-4.1-8b ● 7.2 $0.10 ibm-granite 8.8B — 1
— llama-3.1-nemotron-nano-8b-v1 ● 3.7 — nvidia 8B — —
— gemma-3-12b-it ● 3.1 $0.10 google 12B 128k 5
— gemma-3n-e4b ● 3.1 — google 7.8B — —
— granite-3.3-8b-instruct ● 3.1 $0.25 ibm-granite 8.2B — 1
— molmo2-8b ● 3.1 — allenai 8.7B — —
— olmo-3-7b-think ● 3.1 — allenai 7.3B — —
— phi-4 ● 3.1 $0.14 microsoft 14.7B 16k 3
— gemma-3n-e2b-it ● — — google 5.4B — —
— gemma-2-9b-it ● — — google 9.2B — —
— gigachat3.5-432b-a28b ● — — ai-sage 432B (A28B) — —
— granite-4.0-h-micro ● — $0.11 ibm-granite 3.2B 131k 1
— hy4-preview ● — $2.50 tencent 780B 1024k 1
— laguna-m.1 ● — — poolside 225.8B — —
— laguna-s-2.1 ● — $0.18 poolside 117.6B 1024k 1
— laguna-xs.2 ● — — poolside 33.4B — —
— leanstral-1.5-119b-a6b ● — — mistralai 119B (A6B) — 1
— ling-3.0-flash-fin ● — $0.18 inclusionai 127.5B 256k 2
— llama-3.1-405b-instruct ● — $0.30 meta-llama 405B — 3
— mimo-v2.5 ● — $0.24 xiaomi 310.8B 1050k 9
— nex-n2-mini ● — $0.10 nex-agi 35.1B 256k 1
— olmo-2-32b-instruct ● — — allenai 32B — —
— olmo-3-32b ● — — allenai 32.2B — —
— ornith-1.0-35b ● — — deepreinforce-ai 35B — —
— ornith-1.0-397b ● — — deepreinforce-ai 397B — —
— ornith-1.0-9b ● — — deepreinforce-ai 9B — —
— phi-3.5-mini-instruct ● — $0.52 microsoft 3.8B — 1
— phi-4-multimodal-instruct ● — $0.32 microsoft 5.6B — 1
— qwen-agentworld-35b-a3b ● — — qwen 34.7B (A3.3B) — —
— qwen2.5-omni-7b ● — — qwen 10.7B — —
— qwen2.5-vl-72b-instruct ● — $0.40 qwen 73.4B 125k 5
— qwen3-14b ● — $0.20 qwen 14.8B 128k 7