Spaces:
Running
Running
| component_id,family,organization,unit,quantity,quantity_provenance,scope,sources,conversion_rates,low_musd,central_musd,high_musd | |
| llama2-published-training,Llama 2,Meta,million_A100_hours,3.3,Published hardware-hours; rental price is assumed,All released sizes; no generic allowance.,https://github.com/meta-llama/llama/blob/main/MODEL_CARD.md,1 | 2 | 3,3.3,6.6,9.899999999999999 | |
| llama31-cumulative,Llama 3 / 3.1 / 3.2 / 3.3,Meta,million_H100_hours,39.3,Published hardware-hours; rental price is assumed,3.1 reported total;3.0 overlap unresolved and excluded here;3.3 repetition excluded.,https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD_VISION.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/MODEL_CARD.md,1.5 | 2.5 | 4,58.949999999999996,98.25,157.2 | |
| llama32-text-increment,Llama 3 / 3.1 / 3.2 / 3.3,Meta,million_H100_hours,0.916,Published hardware-hours; rental price is assumed,Published text total incl86k logit generation; text/table rounding discrepancy retained in note.,https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD_VISION.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/MODEL_CARD.md,1.5 | 2.5 | 4,1.374,2.29,3.664 | |
| llama32-vision-increment,Llama 3 / 3.1 / 3.2 / 3.3,Meta,million_H100_hours,2.02,Published hardware-hours; rental price is assumed,11B/90B vision additional hours; no second Llama3 base.,https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/MODEL_CARD_VISION.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3/MODEL_CARD.md | https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/MODEL_CARD.md,1.5 | 2.5 | 4,3.0300000000000002,5.05,8.08 | |
| llama4-pretraining,Llama 4 Scout + Maverick,Meta,million_H100_hours,7.38,Published hardware-hours; rental price is assumed,Published pretraining only; posttraining unknown.,https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md,1.5 | 2.5 | 4,11.07,18.45,29.52 | |
| qwen3-original-eight-s1,Qwen 3,Alibaba,1e24_FLOP,15.353999999999997,Reported parameters/tokens;6ND proxy unless specified,"All eight models >30T S1; conservative30T slice, not a claim that complete training is30T.",https://huggingface.co/Qwen/Qwen3-235B-A22B | https://arxiv.org/html/2505.09388v1 | https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct | https://qwenlm.github.io/blog/qwen3-coder/ | https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct,0.8 | 1.8 | 4,12.283199999999999,27.637199999999996,61.41599999999999 | |
| qwen3-coder-480b,Qwen 3,Alibaba,1e24_FLOP,1.575,Reported parameters/tokens;6ND proxy unless specified,35B active×7.5T tokens; official blog confirmed.,https://qwenlm.github.io/blog/qwen3-coder/,0.8 | 1.8 | 4,1.26,2.835,6.3 | |
| qwen3-next-80b,Qwen 3,Alibaba,1e24_FLOP,0.27,Reported parameters/tokens;6ND proxy unless specified,3B active×15T; official card.,https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct,0.8 | 1.8 | 4,0.21600000000000003,0.48600000000000004,1.08 | |
| deepseek-v2,DeepSeek V2,DeepSeek,1e24_FLOP,1.0206,Reported parameters/tokens;6ND proxy unless specified,21B active×8.1T.,https://huggingface.co/deepseek-ai/DeepSeek-V2,0.8 | 1.8 | 4,0.81648,1.83708,4.0824 | |
| deepseek-v3-published,DeepSeek V3 / R1 / V3.x,DeepSeek,million_H800_hours,2.788,Published hardware-hours; rental price is assumed,Includes published context/posttraining; no inferred R1 or V3.x budget.,https://huggingface.co/deepseek-ai/DeepSeek-V3-Base | https://arxiv.org/html/2412.19437v2,1 | 2 | 3,2.788,5.576,8.363999999999999 | |
| deepseek-v4-pro,DeepSeek V4 Pro + Flash,DeepSeek,1e24_FLOP,9.702,Reported parameters/tokens;6ND proxy unless specified,49B active×33T.,https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro | https://arxiv.org/html/2606.19348,0.8 | 1.8 | 4,7.7616000000000005,17.4636,38.808 | |
| deepseek-v4-flash,DeepSeek V4 Pro + Flash,DeepSeek,1e24_FLOP,2.496,Reported parameters/tokens;6ND proxy unless specified,13B active×32T.,https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro | https://arxiv.org/html/2606.19348,0.8 | 1.8 | 4,1.9968000000000001,4.4928,9.984 | |
| kimi-k2-base,Kimi K2 / K2.5,Moonshot,1e24_FLOP,2.976,Reported parameters/tokens;6ND proxy unless specified,32B active×15.5T.,https://huggingface.co/moonshotai/Kimi-K2-Base | https://arxiv.org/html/2602.02276,0.8 | 1.8 | 4,2.3808000000000002,5.3568,11.904 | |
| kimi-k25-increment,Kimi K2 / K2.5,Moonshot,1e24_FLOP,2.88,Reported parameters/tokens;6ND proxy unless specified,32B active×15T ADDITIONAL tokens; base not counted again.,https://huggingface.co/moonshotai/Kimi-K2-Base | https://arxiv.org/html/2602.02276,0.8 | 1.8 | 4,2.304,5.184,11.52 | |
| longcat-flash-20t-slice,LongCat Flash + 2.0,Meituan,1e24_FLOP,3.24,Reported parameters/tokens;6ND proxy unless specified,Average27B active×20T documented slice; no assumed extra3T.,https://huggingface.co/meituan-longcat/LongCat-2.0 | https://arxiv.org/html/2509.01322,0.8 | 1.8 | 4,2.5920000000000005,5.832000000000001,12.96 | |
| longcat2-35t-slice,LongCat Flash + 2.0,Meituan,1e24_FLOP,10.08,Reported parameters/tokens;6ND proxy unless specified,48B active×35T documented lower token slice; ASIC efficiency uncertainty not resolved.,https://huggingface.co/meituan-longcat/LongCat-2.0 | https://arxiv.org/html/2509.01322,0.8 | 1.8 | 4,8.064,18.144000000000002,40.32 | |
| ling1t-base,Ling 1T / 2.6,Ant Group,1e24_FLOP,6,Reported parameters/tokens;6ND proxy unless specified,50B active×20T.,https://huggingface.co/inclusionAI/Ling-1T | https://huggingface.co/inclusionAI/Ling-2.6-1T-base,0.8 | 1.8 | 4,4.800000000000001,10.8,24 | |
| ling26-increment,Ling 1T / 2.6,Ant Group,1e24_FLOP,2.88,Reported parameters/tokens;6ND proxy unless specified,50B active×9.6T continuation.,https://huggingface.co/inclusionAI/Ling-1T | https://huggingface.co/inclusionAI/Ling-2.6-1T-base,0.8 | 1.8 | 4,2.304,5.184,11.52 | |
| glm45-and-air,GLM 4.5 / Air,Z.ai,1e24_FLOP,6.0984,Reported parameters/tokens;6ND proxy unless specified,(32+12)B active×23.1T.,https://huggingface.co/zai-org/GLM-4.5-Base,0.8 | 1.8 | 4,4.87872,10.97712,24.3936 | |
| glm5-base,GLM 5 / 5.x,Z.ai,1e24_FLOP,6.84,Reported parameters/tokens;6ND proxy unless specified,40B active×28.5T; no later-version or RL margin.,https://huggingface.co/zai-org/GLM-5,0.8 | 1.8 | 4,5.472,12.312,27.36 | |
| hunyuan-large-reported-flops,Hunyuan Large / Hy3 / Hy4,Tencent,1e24_FLOP,3.49,Reported parameters/tokens;6ND proxy unless specified,Paper architecture-specific9.59ND; retain this rather than cheaper6ND proxy.,https://arxiv.org/abs/2411.02265,0.8 | 1.8 | 4,2.7920000000000003,6.282000000000001,13.96 | |
| nemotron3-ultra,Nemotron 3 / 3.5,NVIDIA,1e24_FLOP,6.6,Reported parameters/tokens;6ND proxy unless specified,55B active×20T; NVFP4 pricing not observed.,https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-Base-BF16 | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-Base-BF16 | https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 | https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16 | https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16,0.8 | 1.8 | 4,5.28,11.879999999999999,26.4 | |
| nemotron3-super,Nemotron 3 / 3.5,NVIDIA,1e24_FLOP,1.8,Reported parameters/tokens;6ND proxy unless specified,12B active×25T trained; unique-corpus size not used.,https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-Base-BF16,0.8 | 1.8 | 4,1.4400000000000002,3.24,7.2 | |
| nemotron3-nano,Nemotron 3 / 3.5,NVIDIA,1e24_FLOP,0.48,Reported parameters/tokens;6ND proxy unless specified,"3.2B active×25T trained, NOT10.6T unique corpus.",https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16,0.8 | 1.8 | 4,0.384,0.864,1.92 | |
| nemotron4-340b,Nemotron 4 / H,NVIDIA,1e24_FLOP,18.36,Reported parameters/tokens;6ND proxy unless specified,340B×9T(8T+1T).,https://huggingface.co/nvidia/Nemotron-4-340B-Base | https://arxiv.org/abs/2504.03624,0.8 | 1.8 | 4,14.688,33.048,73.44 | |
| nemotron-h56b,Nemotron 4 / H,NVIDIA,1e24_FLOP,6.72,Reported parameters/tokens;6ND proxy unless specified,56B×20T; no pruned47B base duplicated.,https://huggingface.co/nvidia/Nemotron-4-340B-Base | https://arxiv.org/abs/2504.03624,0.8 | 1.8 | 4,5.376,12.096,26.88 | |
| nemotron-h8b,Nemotron 4 / H,NVIDIA,1e24_FLOP,0.72,Reported parameters/tokens;6ND proxy unless specified,8B×15T; omitted in V1.,https://huggingface.co/nvidia/Nemotron-4-340B-Base | https://arxiv.org/abs/2504.03624,0.8 | 1.8 | 4,0.576,1.296,2.88 | |
| minimax-text01,MiniMax Text-01 / M1,MiniMax,1e24_FLOP,3.13956,Reported parameters/tokens;6ND proxy unless specified,45.9B×11.4T(7.2+3.2+1).,https://huggingface.co/MiniMaxAI/MiniMax-Text-01 | https://arxiv.org/html/2506.13585 | https://arxiv.org/html/2501.08313,0.8 | 1.8 | 4,2.511648,5.651208,12.55824 | |
| minimax-m1-cpt,MiniMax Text-01 / M1,MiniMax,1e24_FLOP,2.0655,Reported parameters/tokens;6ND proxy unless specified,45.9B×7.5T incremental.,https://huggingface.co/MiniMaxAI/MiniMax-Text-01 | https://arxiv.org/html/2506.13585 | https://arxiv.org/html/2501.08313,0.8 | 1.8 | 4,1.6524,3.7179,8.262 | |
| minimax-m1-rl-published,MiniMax Text-01 / M1,MiniMax,million_USD_reported,0.53,Published rental-equivalent cost,Separate reported RL estimate; not invoice. Reporting rate embedded.,https://huggingface.co/MiniMaxAI/MiniMax-Text-01 | https://arxiv.org/html/2506.13585 | https://arxiv.org/html/2501.08313,1 | 1 | 1,0.53,0.53,0.53 | |
| falcon-family,Falcon 7B / 40B / 180B,TII,1e24_FLOP,4.0614,180B reportedPF-days; smaller6ND,180B3.7584e24 +40B.24 +7B.063. Older-generation conversion prices.,https://arxiv.org/html/2311.16867 | https://huggingface.co/tiiuae/falcon-180B,1.5 | 3 | 6,6.0921,12.1842,24.3684 | |
| gemma2-three-sizes,Gemma 2,Google,1e24_FLOP,2.562,Reported parameters/tokens;6ND proxy unless specified,27B13T +9B8T +2B2T; teacher forward cost unpriced.,https://arxiv.org/html/2408.00118,0.8 | 1.8 | 4,2.0496,4.6116,10.248 | |
| gemma3-four-sizes,Gemma 3,Google,1e24_FLOP,3.24,Reported parameters/tokens;6ND proxy unless specified,27B14T +12B12T +4B4T +1B2T; no3n/270M extrapolation.,https://arxiv.org/html/2503.19786,0.8 | 1.8 | 4,2.5920000000000005,5.832000000000001,12.96 | |
| gptoss-two-sizes,gpt-oss 120B / 20B,OpenAI,million_H100_hours,2.31,Published hardware-hours; rental price is assumed,"120B2.1M;20B almost10×less, approximated. No extra posttraining margin.",https://arxiv.org/html/2508.10925,1.5 | 2.5 | 4,3.465,5.775,9.24 | |
| olmo2-three-sizes,OLMo 2,Ai2,1e24_FLOP,1.71,Reported parameters/tokens;6ND proxy unless specified,7B4T+13B5T+32B6T.,https://huggingface.co/allenai/OLMo-2-0325-32B,0.8 | 1.8 | 4,1.368,3.078,6.84 | |
| granite3-dense,Granite 3 dense,IBM,1e24_FLOP,0.72,Reported parameters/tokens;6ND proxy unless specified,"8B+2B,12T each; other generations not included.",https://huggingface.co/ibm-granite/granite-3.0-8b-base,0.8 | 1.8 | 4,0.576,1.296,2.88 | |
| phi4-14b,Phi 4,Microsoft,1e24_FLOP,0.8232,Reported parameters/tokens;6ND proxy unless specified,14B9.8T; Mini not included.,https://huggingface.co/microsoft/phi-4,0.8 | 1.8 | 4,0.65856,1.4817600000000002,3.2928 | |
| trinity-large-final,Trinity Large,Arcee / Prime Intellect,1e24_FLOP,1.326,Reported parameters/tokens;6ND proxy unless specified,13B active×17T final;10T checkpoint not added.,https://arxiv.org/html/2602.17004 | https://huggingface.co/arcee-ai/Trinity-Large-TrueBase,0.8 | 1.8 | 4,1.0608000000000002,2.3868,5.304 | |
| inkling-large,Inkling / Small,Thinking Machines,1e24_FLOP,11.07,Reported parameters/tokens;6ND proxy unless specified,41B active×45T from launch post; Small excluded.,https://thinkingmachines.ai/news/introducing-inkling/,0.8 | 1.8 | 4,8.856,19.926000000000002,44.28 | |
| mimo-three-bases,MiMo 7B / V2 Flash / V2.5 Pro,Xiaomi,1e24_FLOP,10.284,Reported parameters/tokens;6ND proxy unless specified,7B25T +15B27T +42B27T.,https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash | https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro | https://arxiv.org/abs/2505.07608,0.8 | 1.8 | 4,8.227200000000002,18.511200000000002,41.136 | |
| telechat2-115b,TeleChat 1 / 2,China Telecom,1e24_FLOP,6.9,Reported parameters/tokens;6ND proxy unless specified,115B10T primary card; smaller models omitted from documented slice.,https://huggingface.co/Tele-AI/TeleChat2-115B | https://github.com/Tele-AI/TeleChat2,0.8 | 1.8 | 4,5.5200000000000005,12.420000000000002,27.6 | |
| apertus70-6nd,Apertus 8B / 70B,Swiss AI,1e24_FLOP,6.3,Reported parameters/tokens;6ND proxy unless specified,70B15T primary card;8B and reported6.74e24 secondary value not assumed here.,https://huggingface.co/swiss-ai/Apertus-70B-2509 | https://github.com/swiss-ai/apertus-tech-report/blob/main/Apertus_Tech_Report.pdf,0.8 | 1.8 | 4,5.04,11.34,25.2 | |
| solar-open2-250b,Solar Open 100B / Open2 250B,Upstage,million_B200_hours,2,Published hardware-hours; rental price is assumed,Primary card2M GPU-hours; divergent Epoch$32M not added.,https://huggingface.co/upstage/Solar-Open2-250B | https://huggingface.co/upstage/Solar-Open-100B | https://epoch.ai/data/ai-models,3 | 5 | 8,6,10,16 | |
| solar-open100b,Solar Open 100B / Open2 250B,Upstage,1e24_FLOP,1.4184,Reported parameters/tokens;6ND proxy unless specified,12B active×19.7T.,https://huggingface.co/upstage/Solar-Open2-250B | https://huggingface.co/upstage/Solar-Open-100B | https://epoch.ai/data/ai-models,0.8 | 1.8 | 4,1.1347200000000002,2.5531200000000003,5.6736 | |
| dbrx-base,DBRX,Databricks,1e24_FLOP,2.592,Reported parameters/tokens;6ND proxy unless specified,36B active×12T.,https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm | https://huggingface.co/databricks/dbrx-base,0.8 | 1.8 | 4,2.0736000000000003,4.6656,10.368 | |