AI Hippo

Hottest open-source projects in the last 24 hours

meta-llama/Llama-3.3-70B-Instruct · +1.2k ⭐ / 24h deepseek-ai/DeepSeek-V3 · HF 下载 +18% Qwen/Qwen2.5-72B-Instruct · Stars +890 open-webui/open-webui · +420 ⭐ hiyouga/LLaMA-Factory · +310 ⭐ unslothai/unsloth · Hugging Face 周下载 +22% meta-llama/Llama-3.3-70B-Instruct · +1.2k ⭐ / 24h deepseek-ai/DeepSeek-V3 · HF 下载 +18% Qwen/Qwen2.5-72B-Instruct · Stars +890 open-webui/open-webui · +420 ⭐ hiyouga/LLaMA-Factory · +310 ⭐ unslothai/unsloth · Hugging Face 周下载 +22%

GitHub + HF

Tendencias OSS en GitHub

Además de los rankings, miramos cambios estructurales en el ecosistema open source en GitHub: actividad, atención y calor comunitario por dominio y tipo de proyecto.

Hot models

  • Qwen2.5-72B-Instruct

    Alibaba

    +4.2%
    VRAM
    ≥48GB
    License
    Apache-2.0
    HF
    Model card

    Apache-2.0

  • Llama-3.3-70B

    Meta

    +2.8%
    VRAM
    ≥40GB
    License
    Llama 3.3
    HF
    Model card

    Llama 3.3 Community

  • DeepSeek-V3

    DeepSeek

    +6.1%
    VRAM
    ≥80GB
    License
    MIT
    HF
    Model card

    MIT

  • Flux.1-dev

    Black Forest Labs

    +3.5%
    VRAM
    ≥24GB
    License
    FLUX.1 [dev] Non-Commercial
    HF
    Model card

    非商业许可

Grouped repository comparison

Each section lists repositories of the same class (e.g., Agent: OpenClaw, NanoBot, PicoClaw). Column meanings are consistent across groups. “Commits (30d)” follows the scraper’s rolling window.

Snapshot:

Public formula (within each group): Score = 100 × [0.30·Stars + 0.15·Forks + 0.30·Commits30d + 0.15·Contributors + 0.05·(1-Issues) + 0.05·(1-PRs)], after per-group min-max normalization.

Agent

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 openclaw/openclaw 82,6 344.000 68.000 920 2100 4200 380
2 langchain-ai/langchain 47 106.000 16.800 412 4100 2100 98
3 Significant-Gravitas/AutoGPT 37,7 176.000 44.000 42 1200 1200 52
4 crewAIInc/crewAI 19 32.000 4100 124 890 420 36
5 All-Hands-AI/OpenHands 18 52.000 5800 132 460 1400 72
6 hkuds/nanobot 17,2 37.400 4100 156 180 890 42
7 langchain-ai/langgraph 17 11.800 1900 198 420 680 45
8 webx-trend-demo/agent-slot-20 15 3835 495 119 843 847 21
9 qwibitai/nanoclaw 15 26.600 2800 112 95 410 28
10 sipeed/picoclaw 14,1 25.700 2100 88 64 320 18
11 webx-trend-demo/agent-slot-19 13,3 3813 473 97 821 825 84
12 webx-trend-demo/agent-slot-18 13,3 3812 472 96 820 824 83
13 webx-trend-demo/agent-slot-17 13,3 3811 471 95 819 823 82
14 webx-trend-demo/agent-slot-16 13,3 3810 470 94 818 822 81
15 webx-trend-demo/agent-slot-15 13,2 3809 469 93 817 821 80
16 webx-trend-demo/agent-slot-14 13,2 3808 468 92 816 820 79
17 webx-trend-demo/agent-slot-13 13,2 3807 467 91 815 819 78
18 webx-trend-demo/agent-slot-12 13,2 3806 466 90 814 818 77
19 webx-trend-demo/agent-slot-11 13,1 3805 465 89 813 817 76
20 Pythagora-io/gpt-pilot 13 33.000 3000 36 130 520 24

Multi-agent

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 browser-use/browser-use 82,6 62.000 7200 210 210 890 48
2 microsoft/autogen 63,5 42.000 6100 86 512 540 34
3 geekan/MetaGPT 62,9 51.000 6400 78 390 760 30
4 OpenInterpreter/open-interpreter 54 58.000 6200 38 210 890 28
5 webx-trend-demo/multi_agent-slot-20 33,5 13.745 2005 109 53 57 11
6 ag2ai/ag2 30,4 14.000 1700 74 240 390 26
7 SWE-agent/SWE-agent 27,2 15.000 1900 64 120 280 22
8 webx-trend-demo/multi_agent-slot-19 25 13.723 1983 87 31 35 74
9 webx-trend-demo/multi_agent-slot-18 24,9 13.722 1982 86 30 34 73
10 webx-trend-demo/multi_agent-slot-17 24,8 13.721 1981 85 29 33 72
11 webx-trend-demo/multi_agent-slot-16 24,7 13.720 1980 84 28 32 71
12 microsoft/JARVIS 24,6 22.000 2700 10 72 210 5
13 webx-trend-demo/multi_agent-slot-15 24,6 13.719 1979 83 27 31 70
14 webx-trend-demo/multi_agent-slot-14 24,5 13.718 1978 82 26 30 69
15 webx-trend-demo/multi_agent-slot-13 24,4 13.717 1977 81 25 29 68
16 webx-trend-demo/multi_agent-slot-12 24,3 13.716 1976 80 24 28 67
17 webx-trend-demo/multi_agent-slot-11 24,2 13.715 1975 79 23 27 66
18 langroid-ai/langroid 17,2 4200 480 52 85 120 12
19 aorwall/moa 16,8 11.000 1200 18 68 140 9
20 microsoft/TaskWeaver 16,6 7600 860 28 96 150 8

Inference

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 vllm-project/vllm 76,4 48.000 8200 380 920 2100 180
2 ggerganov/llama.cpp 70,6 72.100 10.300 142 684 890 62
3 microsoft/onnxruntime 39,8 18.000 3200 165 890 1200 120
4 apache/tvm 30,2 14.000 3200 88 760 980 85
5 sgl-project/sglang 27 12.000 1500 210 210 890 95
6 Tencent/ncnn 25,6 21.000 4800 48 320 680 42
7 pytorch/executorch 22,2 12.000 1100 136 190 420 36
8 mlc-ai/mlc-llm 21,5 22.000 1900 58 210 410 21
9 webx-trend-demo/inference-slot-20 16,9 1527 987 91 335 339 23
10 NVIDIA/TensorRT-LLM 15,5 9800 1800 72 210 890 42
11 internlm/lmdeploy 15,2 7600 920 86 150 320 30
12 webx-trend-demo/inference-slot-19 15,1 1505 965 69 313 317 1
13 webx-trend-demo/inference-slot-18 12,7 1504 964 68 312 316 85
14 webx-trend-demo/inference-slot-17 12,6 1503 963 67 311 315 84
15 webx-trend-demo/inference-slot-16 12,5 1502 962 66 310 314 83
16 webx-trend-demo/inference-slot-15 12,4 1501 961 65 309 313 82
17 webx-trend-demo/inference-slot-14 12,3 1500 960 64 308 312 81
18 webx-trend-demo/inference-slot-13 12,3 1499 959 63 307 311 80
19 webx-trend-demo/inference-slot-12 12,2 1498 958 62 306 310 79
20 webx-trend-demo/inference-slot-11 12,1 1497 957 61 305 309 78

LLM library

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 huggingface/transformers 78,6 142.000 28.300 198 2950 1120 156
2 google-gemini/gemini-cli 39,1 18.000 2100 320 210 890 48
3 BerriAI/litellm 30,5 17.000 2100 220 260 640 52
4 unslothai/unsloth 27,6 42.000 3200 165 85 1200 62
5 OpenBMB/MiniCPM 17,2 18.000 1400 62 130 320 16
6 ml-explore/mlx 15,7 21.000 1500 42 120 280 28
7 QwenLM/Qwen 14,9 17.000 1300 40 110 280 15
8 webx-trend-demo/llm_lib-slot-20 14,6 6989 849 73 497 501 75
9 webx-trend-demo/llm_lib-slot-19 13,2 6967 827 51 475 479 53
10 webx-trend-demo/llm_lib-slot-18 13,1 6966 826 50 474 478 52
11 webx-trend-demo/llm_lib-slot-17 13,1 6965 825 49 473 477 51
12 webx-trend-demo/llm_lib-slot-16 13 6964 824 48 472 476 50
13 webx-trend-demo/llm_lib-slot-15 12,9 6963 823 47 471 475 49
14 webx-trend-demo/llm_lib-slot-14 12,9 6962 822 46 470 474 48
15 webx-trend-demo/llm_lib-slot-13 12,8 6961 821 45 469 473 47
16 Lightning-AI/litgpt 12,8 13.000 1400 26 170 310 20
17 webx-trend-demo/llm_lib-slot-12 12,7 6960 820 44 468 472 46
18 webx-trend-demo/llm_lib-slot-11 12,7 6959 819 43 467 471 45
19 LLaVA-VL/LLaVA-NeXT 12,5 15.000 1200 18 90 240 12
20 mistralai/mistral-inference 10,5 8900 920 8 42 120 8

Training framework

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 tensorflow/tensorflow 73,2 189.000 89.000 210 4200 2100 420
2 pytorch/pytorch 63,7 92.000 26.500 620 5120 7200 820
3 keras-team/keras 30,3 63.000 20.000 94 1500 520 96
4 jax-ml/jax 24,2 32.000 3100 180 890 1200 120
5 tinygrad/tinygrad 24 38.000 3900 170 460 880 74
6 PaddlePaddle/Paddle 21,1 24.000 5400 140 980 1600 150
7 mindspore-ai/mindspore 19,2 29.000 5200 88 620 980 63
8 webx-trend-demo/framework-slot-20 15,2 12.146 406 70 554 558 17
9 apache/mxnet 14,8 21.000 6800 12 890 1800 42
10 facebookresearch/xformers 14,3 9000 1000 84 240 460 55
11 openxla/xla 13,8 3600 420 92 210 380 44
12 webx-trend-demo/framework-slot-19 13,6 12.124 384 48 532 536 80
13 webx-trend-demo/framework-slot-18 13,6 12.123 383 47 531 535 79
14 webx-trend-demo/framework-slot-17 13,5 12.122 382 46 530 534 78
15 webx-trend-demo/framework-slot-16 13,5 12.121 381 45 529 533 77
16 webx-trend-demo/framework-slot-15 13,5 12.120 380 44 528 532 76
17 webx-trend-demo/framework-slot-14 13,4 12.119 379 43 527 531 75
18 webx-trend-demo/framework-slot-13 13,4 12.118 378 42 526 530 74
19 webx-trend-demo/framework-slot-12 13,3 12.117 377 41 525 529 73
20 webx-trend-demo/framework-slot-11 13,3 12.116 376 40 524 528 72

Serving

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 ollama/ollama 72,1 138.000 11.200 124 420 1800 72
2 ray-project/ray 61,5 37.000 6500 210 2300 5200 420
3 mudler/LocalAI 32,9 32.000 2800 95 210 890 48
4 lm-sys/FastChat 25,7 39.000 4600 16 240 780 18
5 skypilot-org/skypilot 23 6800 520 84 125 210 24
6 huggingface/text-generation-inference 22,5 12.000 1500 64 180 420 36
7 KServe/KServe 21,5 4200 1400 66 320 520 62
8 vllm-project/aibrix 19,3 1300 120 72 60 90 10
9 triton-inference-server/server 18,8 8900 2100 42 120 680 28
10 webx-trend-demo/serving-slot-20 16,3 3476 136 40 284 288 12
11 bentoml/OpenLLM 15 11.000 840 22 105 230 14
12 webx-trend-demo/serving-slot-19 12,1 3454 114 18 262 266 75
13 webx-trend-demo/serving-slot-18 11,9 3453 113 17 261 265 74
14 webx-trend-demo/serving-slot-17 11,8 3452 112 16 260 264 73
15 webx-trend-demo/serving-slot-16 11,6 3451 111 15 259 263 72
16 webx-trend-demo/serving-slot-15 11,5 3450 110 14 258 262 71
17 webx-trend-demo/serving-slot-14 11,3 3449 109 13 257 261 70
18 webx-trend-demo/serving-slot-13 11,2 3448 108 12 256 260 69
19 webx-trend-demo/serving-slot-12 11,1 3447 107 11 255 259 68
20 webx-trend-demo/serving-slot-11 10,9 3446 106 10 254 258 67

Evaluation

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 openai/evals 64,8 15.000 2500 14 310 640 24
2 explodinggradients/ragas 63,8 9000 950 118 180 410 32
3 confident-ai/deepeval 58,2 9400 780 96 160 330 20
4 webx-trend-demo/eval-slot-19 55,6 9262 322 106 170 174 73
5 webx-trend-demo/eval-slot-18 55,4 9261 321 105 169 173 72
6 webx-trend-demo/eval-slot-17 55,1 9260 320 104 168 172 71
7 webx-trend-demo/eval-slot-16 54,9 9259 319 103 167 171 70
8 webx-trend-demo/eval-slot-15 54,6 9258 318 102 166 170 69
9 webx-trend-demo/eval-slot-14 54,4 9257 317 101 165 169 68
10 webx-trend-demo/eval-slot-13 54,1 9256 316 100 164 168 67
11 webx-trend-demo/eval-slot-12 53,9 9255 315 99 163 167 66
12 webx-trend-demo/eval-slot-11 53,6 9254 314 98 162 166 65
13 EleutherAI/lm-evaluation-harness 53,5 9200 1200 48 320 420 42
14 webx-trend-demo/eval-slot-20 34,8 9284 344 8 192 196 10
15 open-compass/opencompass 32,2 4800 480 52 120 280 18
16 RUC-NLPIR/FlashRAG 31,8 6100 560 40 88 210 11
17 huggingface/lighteval 17,9 2100 180 38 42 120 12
18 stanford-crfm/helm 15,4 2800 320 12 85 180 8
19 lmarena/arena-hard-auto 14,9 2200 240 20 55 96 7
20 prometheus-eval/prometheus-eval 11,6 1900 160 14 42 70 6

RAG / vector search

Rank Repository Score Stars Forks Commits (30d) Contributors Open issues Open PRs
1 langgenius/dify 83,4 82.000 12.000 260 980 3100 160
2 run-llama/llama_index 55,3 42.000 6200 165 890 1200 85
3 FlowiseAI/Flowise 49 39.000 21.000 96 520 1300 84
4 AnythingLLM/AnythingLLM 39,3 41.000 4200 120 340 820 55
5 milvus-io/milvus 33,9 32.000 4200 120 420 2100 95
6 deepset-ai/haystack 26,5 19.800 2600 72 420 470 28
7 qdrant/qdrant 26,4 24.000 1500 95 210 420 42
8 vercel/ai-chatbot 19,6 17.000 3400 54 160 240 22
9 webx-trend-demo/rag-slot-20 17,4 1160 620 84 268 272 31
10 weaviate/weaviate 16,9 12.000 890 64 180 420 36
11 chroma-core/chroma 15,3 18.000 1500 42 120 680 28
12 webx-trend-demo/rag-slot-19 14,7 1138 598 62 246 250 9
13 webx-trend-demo/rag-slot-18 14,6 1137 597 61 245 249 8
14 webx-trend-demo/rag-slot-17 14,4 1136 596 60 244 248 7
15 webx-trend-demo/rag-slot-16 14,3 1135 595 59 243 247 6
16 webx-trend-demo/rag-slot-15 14,2 1134 594 58 242 246 5
17 webx-trend-demo/rag-slot-14 14,1 1133 593 57 241 245 4
18 webx-trend-demo/rag-slot-13 14 1132 592 56 240 244 3
19 webx-trend-demo/rag-slot-12 13,8 1131 591 55 239 243 2
20 webx-trend-demo/rag-slot-11 13,7 1130 590 54 238 242 1

Demo data—not live GitHub. Production: schedule REST/GraphQL ingestion, write JSON, then rebuild.

Métricas de ejemplo

Velocidad de estrellas, cadencia de commits recientes, actividad en issues/PRs, distribución de contribuidores, ritmo de releases—por día/semana/mes para series o mapas de calor.

Ingesta y actualización

GitHub REST/GraphQL, datasets públicos o agregadores como RSSHub; Actions programados para normalizar y escribir JSON o D1, luego páginas estáticas o APIs de solo lectura en el edge.

Tipos y dominios que miramos

  • LLM / pila de inferencia (serving, kernels, cuantización, etc.)
  • Agentes, multi‑agente, uso de herramientas y orquestación
  • Evaluaciones, benchmarks, datasets y tablas reproducibles
  • Entrenamiento, fine‑tuning, alineación y toolchains de ingeniería
  • Despliegue local, servicios de inferencia y edge/endpoint
  • Datos, RAG, recuperación y ecosistemas de vector stores

Common misreadings

  • Equating stars with model quality—hype may diverge from benchmark strength.
  • Relying on short windows only—spikes may reflect campaigns; longer commit curves carry more signal.
  • Ignoring issues/PRs—healthy projects often show dense maintenance beyond stars.
  • Mixing personal and organization repositories—governance and release cadence may differ materially.
  • Treating GitHub as the entire ecosystem—important work may be closed-source or hosted elsewhere.

Más lectura

Volver al inicio