⚡ AI TREND FLASH

Scores are signals, not verdicts. This tracker surfaces velocity, transparency, and local compatibility in open models—so you can choose what fits your hardware, your use case, and your privacy boundary.

🦊 How to read this table

Public leaderboards are snapshots, not guarantees. Many benchmarks suffer from data contamination or reward verbosity over accuracy. Use this table to spot momentum, then verify models in your own environment.

Benchmark scores reflect specific test conditions. Treat them as a starting point, not proof of real-world performance.
Size & architecture often matter more than rank. A well-tuned 7B model will outperform a bloated base model on consumer hardware.
License clarity is critical. "Open weights" ≠ open usage. Look for Apache 2.0, MIT, or permissive commercial terms.
Local-first is the new baseline. Models that run air-gapped, leave no telemetry, and fit in ≤16GB VRAM give you true digital sovereignty.

⚡ Local-First Quick Reference

Estimate VRAM needs before downloading. INT4 quantization cuts memory footprint by ~75% with minimal quality loss.

VRAM Rule	≈ (Parameters × Precision) ÷ 8
Entry (12GB)	7B–8B → direct response, light tasks
Recommended (24GB)	Up to 34B → 30–50 tok/s, coding/reasoning
Workstation (≥48GB)	70B+ → production throughput, multilingual

Rank	Model	Benchmark	Type	Size
1	0-hero/Matter-0.2-7B-DPO	8.91	💬 MistralForCausalLM	7.242B
2	01-ai/Yi-1.5-34B	25.65	🟢 LlamaForCausalLM	34.389B
3	01-ai/Yi-1.5-34B-32K	26.73	🟢 LlamaForCausalLM	34.389B
4	01-ai/Yi-1.5-34B-Chat	33.36	💬 LlamaForCausalLM	34.389B
5	01-ai/Yi-1.5-34B-Chat-16K	29.4	💬 LlamaForCausalLM	34.389B
6	01-ai/Yi-1.5-6B	16.75	🟢 LlamaForCausalLM	6.061B
7	01-ai/Yi-1.5-6B-Chat	22.78	💬 LlamaForCausalLM	6.061B
8	01-ai/Yi-1.5-9B	22.15	🟢 LlamaForCausalLM	8.829B
9	01-ai/Yi-1.5-9B-32K	19.81	🟢 LlamaForCausalLM	8.829B
10	01-ai/Yi-1.5-9B-Chat	29.53	💬 LlamaForCausalLM	8.829B
11	01-ai/Yi-1.5-9B-Chat-16K	23.77	💬 LlamaForCausalLM	8.829B
12	01-ai/Yi-34B	22.37	🟢 LlamaForCausalLM	34.389B
13	01-ai/Yi-34B-200K	20.01	🟢 LlamaForCausalLM	34.389B
14	01-ai/Yi-34B-Chat	24.23	💬 LlamaForCausalLM	34.389B
15	01-ai/Yi-6B	13.61	🟢 LlamaForCausalLM	6.061B
16	01-ai/Yi-6B-200K	12.0	🟢 LlamaForCausalLM	6.061B
17	01-ai/Yi-6B-Chat	14.12	💬 LlamaForCausalLM	6.061B
18	01-ai/Yi-9B	17.81	🟢 LlamaForCausalLM	8.829B
19	01-ai/Yi-9B-200K	17.73	🟢 LlamaForCausalLM	8.829B
20	01-ai/Yi-Coder-9B-Chat	16.99	🔶 LlamaForCausalLM	8.829B
21	1-800-LLMs/Qwen-2.5-14B-Hindi	36.27	🔶 Qwen2ForCausalLM	14.77B
22	1-800-LLMs/Qwen-2.5-14B-Hindi-Custom-Instruct	31.02	🔶 Qwen2ForCausalLM	14.77B
23	1024m/PHI-4-Hindi	27.49	🔶 LlamaForCausalLM	14.66B
24	1024m/QWEN-14B-B100	41.92	🔶 Qwen2ForCausalLM	14.77B
25	152334H/miqu-1-70b-sf	29.1	🔶 LlamaForCausalLM	68.977B
26	1TuanPham/T-VisStar-7B-v0.1	19.14	💬 MistralForCausalLM	7.294B
27	1TuanPham/T-VisStar-v0.1	19.14	🔶 MistralForCausalLM	7.294B
28	3rd-Degree-Burn/L-3.1-Science-Writer-8B	21.09	🔶 LlamaForCausalLM	8.03B
29	3rd-Degree-Burn/Llama-3.1-8B-Squareroot	11.22	🤝 LlamaForCausalLM	8.03B
30	3rd-Degree-Burn/Llama-3.1-8B-Squareroot-v1	8.04	🤝 LlamaForCausalLM	8.03B
31	3rd-Degree-Burn/Llama-Squared-8B	12.43	🔶 LlamaForCausalLM	8.03B
32	4season/final_model_test_v2	23.09	🔶 LlamaForCausalLM	21.421B
33	AALF/FuseChat-Llama-3.1-8B-Instruct-preview	28.57	💬 LlamaForCausalLM	8.03B
34	AALF/FuseChat-Llama-3.1-8B-SFT-preview	29.23	🔶 LlamaForCausalLM	8.03B
35	AALF/gemma-2-27b-it-SimPO-37K	9.51	💬 Gemma2ForCausalLM	27.227B
36	AALF/gemma-2-27b-it-SimPO-37K-100steps	10.25	💬 Gemma2ForCausalLM	27.227B
37	AELLM/gemma-2-aeria-infinity-9b	31.92	🤝 Gemma2ForCausalLM	9.242B
38	AELLM/gemma-2-lyco-infinity-9b	30.05	💬 Gemma2ForCausalLM	10.159B
39	AGI-0/Art-v0-3B	12.13	🔶 Qwen2ForCausalLM	3.086B
40	AGI-0/Artificium-llama3.1-8B-001	19.49	🔶 LlamaForCausalLM	8.03B
41	AGI-0/smartllama3.1-8B-001	20.42	🔶 LlamaForCausalLM	8.03B
42	AI-MO/NuminaMath-7B-CoT	16.12	🔶 LlamaForCausalLM	6.91B
43	AI-MO/NuminaMath-7B-TIR	14.18	🔶 LlamaForCausalLM	6.91B
44	AI-Sweden-Models/Llama-3-8B-instruct	14.34	🔶 LlamaForCausalLM	8.03B
45	AI-Sweden-Models/gpt-sw3-40b	4.87	🟢 GPT2LMHeadModel	39.927B
46	AI4free/Dhanishtha	11.25	🟩 Qwen2ForCausalLM	1.777B
47	AI4free/t2	11.33	🤝 Qwen2ForCausalLM	7.613B
48	AIDC-AI/Marco-o1	27.64	💬 Qwen2ForCausalLM	7.616B
49	Aashraf995/Creative-7B-nerd	29.98	🤝 Qwen2ForCausalLM	7.616B
50	Aashraf995/Gemma-Evo-10B	34.33	🤝 Gemma2ForCausalLM	10.159B
51	Aashraf995/Qwen-Evo-7B	30.28	🤝 Qwen2ForCausalLM	7.616B
52	Aashraf995/QwenStock-14B	37.13	🤝 Qwen2ForCausalLM	14.766B
53	AbacusResearch/Jallabi-34B	26.19	🤝 LlamaForCausalLM	34.389B
54	Ahdoot/StructuredThinker-v0.3-MoreStructure	23.92	🤝 Qwen2ForCausalLM	3.397B
55	Ahdoot/Test_StealthThinker	22.07	🔶 Qwen2ForCausalLM	3.086B
56	AicoresSecurity/Cybernet-Sec-3B-R1-V0	20.58	🔶 LlamaForCausalLM	3.213B
57	AicoresSecurity/Cybernet-Sec-3B-R1-V0-Coder	22.93	🔶 LlamaForCausalLM	3.213B
58	AicoresSecurity/Cybernet-Sec-3B-R1-V1	20.0	🔶 LlamaForCausalLM	3.213B
59	AicoresSecurity/Cybernet-Sec-3B-R1-V1.1	22.64	🔶 LlamaForCausalLM	3.213B
60	Alepach/notHumpback-M0	5.14	💬 LlamaForCausalLM	3.213B
61	Alepach/notHumpback-M1	4.78	💬 LlamaForCausalLM	3.213B
62	Alepach/notHumpback-M1-v2	5.21	💬 LlamaForCausalLM	3.213B
63	Alibaba-NLP/gte-Qwen2-7B-instruct	13.83	🔶 Qwen2ForCausalLM	7.613B
64	Alsebay/Qwen2.5-7B-test-novelist	27.17	🔶 Qwen2ForCausalLM	7.616B
65	Amaorynho/BBAI2006	3.46	🤝 Qwen2ForCausalLM	1.09B
66	Amaorynho/BBAI270V4	4.55	🤝 Qwen2ForCausalLM	7.616B
67	Amaorynho/BBAIIFEV1	30.58	🤝 LlamaForCausalLM	8.03B
68	Amaorynho/BBAI_375	3.46	🤝 Qwen2ForCausalLM	1.09B
69	Amu/t1-1.5B	12.14	🔶 Qwen2ForCausalLM	1.777B
70	Amu/t1-3B	11.16	💬 Qwen2ForCausalLM	3.397B
71	ArliAI/ArliAI-RPMax-12B-v1.1	20.98	🔶 MistralForCausalLM	12.248B
72	ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.1	23.94	🔶 LlamaForCausalLM	8.03B
73	Arthur-LAGACHERIE/Precis-1B-Instruct	8.85	🔶 LlamaForCausalLM	1.236B
74	Artples/L-MChat-7b	21.24	🤝 MistralForCausalLM	7.242B
75	Artples/L-MChat-Small	15.23	🤝 PhiForCausalLM	2.78B
76	Aryanne/QwentileSwap	43.92	🤝 Qwen2ForCausalLM	32.764B
77	Aryanne/SHBA	29.88	🤝 LlamaForCausalLM	8.03B
78	Aryanne/SuperHeart	25.56	🤝 LlamaForCausalLM	8.03B
79	AtAndDev/Qwen2.5-1.5B-continuous-learnt	16.52	🔶 Qwen2ForCausalLM	1.544B
80	AtAndDev/Qwen2.5-1.5B-continuous-learnt	17.48	💬 Qwen2ForCausalLM	1.544B
81	Ateron/Glowing-Forest-12B	22.61	🤝 MistralForCausalLM	12.248B
82	Ateron/Lotus-Magpic	25.5	🤝 MistralForCausalLM	12.248B
83	Ateron/Way_of_MagPicaro	20.63	🤝 MistralForCausalLM	12.248B
84	AuraIndustries/Aura-4B	16.06	💬 LlamaForCausalLM	4.513B
85	AuraIndustries/Aura-8B	27.36	💬 LlamaForCausalLM	8.03B
86	AuraIndustries/Aura-MoE-2x4B	16.8	💬 MixtralForCausalLM	7.231B
87	AuraIndustries/Aura-MoE-2x4B-v2	17.52	💬 MixtralForCausalLM	7.231B
88	Aurel9/testmerge-7b	20.97	🤝 MistralForCausalLM	7.242B
89	Ayush-Singh/Llama1B-sft-2	3.17	🔶 LlamaForCausalLM	1.236B
90	Azure99/Blossom-V6-14B	32.81	💬 Qwen2ForCausalLM	14.77B
91	Azure99/Blossom-V6-7B	31.05	💬 Qwen2ForCausalLM	7.616B
92	Azure99/blossom-v5-32b	27.72	💬 Qwen2ForCausalLM	32.512B
93	Azure99/blossom-v5-llama3-8b	14.6	💬 LlamaForCausalLM	8.03B
94	Azure99/blossom-v5.1-34b	30.3	💬 LlamaForCausalLM	34.389B
95	Azure99/blossom-v5.1-9b	26.47	💬 LlamaForCausalLM	8.829B
96	BAAI/Gemma2-9B-IT-Simpo-Infinity-Preference	22.61	🔶 Gemma2ForCausalLM	9.242B
97	BAAI/Infinity-Instruct-3M-0613-Llama3-70B	35.58	🔶 LlamaForCausalLM	70.554B
98	BAAI/Infinity-Instruct-3M-0613-Mistral-7B	22.29	🔶 MistralForCausalLM	7.242B
99	BAAI/Infinity-Instruct-3M-0625-Llama3-70B	36.91	🔶 LlamaForCausalLM	70.554B
100	BAAI/Infinity-Instruct-3M-0625-Llama3-8B	22.06	💬 LlamaForCausalLM	8.03B