Use cases › Capability
Work across languages
Models that are trained and evaluated on ten or more languages, for non-English and cross-lingual applications.
Benchmarks that represent this job 12
A model's row on one of these third-party boards is what earns the measured label; the same benchmark name in the model's own card earns author-reported.
Global-MMLUMMMLUFLORESMGSMBelebeleINCLUDEMMLU-ProXXCOPAC-EvalCMMLUPolyMATHWMT24++
Models 202
Narrow by what you can run and pay for. Rows are ordered measured → author-reported → structural, then by watch-list score; click a column to re-sort. Every chip links to its source.
| Model | Evidence | Params | Context | Price from | Runs on | License | Providers | Measured latency |
|---|---|---|---|---|---|---|---|---|
| gemma-4-26B-A4B-it google | author-reported Author reports MMMLU 86.3%decision-model estimate p=0.99 | 25.8B (4B active) | 256K | $0.00 | Single high-end GPU | Apache 2.0 | 6 | 0.9 s · 51 tok/s |
| gemma-4-31B-it google | author-reported Author reports MMMLU 88.4%decision-model estimate p=1.00 | 31.3B | 256K | $0.00 | Single high-end GPU | Apache 2.0 | 5 | — |
| DeepSeek-V4.1-Flash deepseek-ai | author-reported Author reports C-Eval (EM) 92.1Author reports MGSM (EM) 80.2 | 763B | 1M | $0.02 | Cluster | MIT | 8 | — |
| Qwen3.5-9B Qwen | author-reported Author reports C-Eval 88.2decision-model estimate p=1.00 | 9.65B | 256K | $0.08 | Single GPU | Apache 2.0 | 6 | 2.8 s · 117 tok/s |
| DeepSeek-V4-Pro deepseek-ai | author-reported Author reports MMMLU (EM) 90.3Author reports C-Eval (EM) 93.1 +1 | 1.60T | 1M | $0.21 | Cluster | MIT | 5 | — |
| DeepSeek-R1 deepseek-ai | author-reported Author reports C-Eval (EM) 91.8 | 684B | 160K | $0.70 | Cluster | MIT | 3 | — |
| Qwen3.5-397B-A17B Qwen | author-reported Author reports C-Eval 93.0decision-model estimate p=1.00 | 403B (17B active) | 256K | $0.39 | Cluster | Apache 2.0 | 7 | — |
| DeepSeek-V4-Flash deepseek-ai | author-reported Author reports MMMLU (EM) 88.8Author reports C-Eval (EM) 92.1 +1 | 291B | 1M | $0.03 | Cluster | MIT | 4 | — |
| Qwen3.5-35B-A3B Qwen | author-reported Author reports C-Eval 90.2decision-model estimate p=1.00 | 36.0B (3B active) | 256K | $0.07 | Multi-GPU node | Apache 2.0 | 4 | — |
| Qwen3.5-27B Qwen | author-reported Author reports C-Eval 90.5decision-model estimate p=1.00 | 27.8B | 256K | $0.20 | Single high-end GPU | Apache 2.0 | 4 | — |
| Qwen3.5-4B Qwen | author-reported Author reports C-Eval 85.1decision-model estimate p=1.00 | 4.66B | 256K | — | Single GPU | Apache 2.0 | 2 | — |
| DeepSeek-V3 deepseek-ai | author-reported Author reports MGSM (EM) 79.8Author reports C-Eval (Acc.) 90.1 +1 | 684B | 160K | $0.26 | Cluster | Not declared review | 2 | — |
| gemma-4-E4B-it google | author-reported Author reports MMMLU 76.6%decision-model estimate p=1.00 | 8.00B | 128K | — | Single GPU | Apache 2.0 | 1 | — |
| phi-4 microsoft | author-reported Author reports MGSM 80.6 | 14.7B | 16K | $0.06 | Single high-end GPU | MIT | 3 | — |
| Qwen3.5-122B-A10B Qwen | author-reported Author reports C-Eval 91.9decision-model estimate p=1.00 | 125B (10B active) | 256K | $0.26 | Multi-GPU node | Apache 2.0 | 3 | — |
| gemma-4-E2B-it google | author-reported Author reports MMMLU 67.4%decision-model estimate p=0.99 | 5.12B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen2-7B-Instruct Qwen | author-reported Author reports C-Eval 77.2 | 7.62B | 32K | — | Single GPU | Apache 2.0 | 2 | — |
| Qwen3.5-2B Qwen | author-reported Author reports C-Eval 73.2Author reports MMMLU 56.9 +1 | 2.27B | 256K | — | Single GPU | Apache 2.0 | 2 | — |
| Qwen3-4B-Thinking-2507 Qwen | author-reported Author reports MMLU-ProX 64.2Author reports INCLUDE 64.4 +1 | 4.02B | 256K | $0.01 | Single GPU | Apache 2.0 | 2 | — |
| Qwen3.5-0.8B Qwen | author-reported Author reports C-Eval 50.5Author reports MMMLU 34.1 +1 | 873M | 256K | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-4-12B-it google | author-reported Author reports MMMLU 83.4%decision-model estimate p=0.99 | 12.0B | 256K | — | Single high-end GPU | Apache 2.0 | 0 | — |
| Inkling thinkingmachines | author-reported Author reports Global-MMLU-Lite 88.7% | 952B | 1M | $0.00 | Cluster | Apache 2.0 | 4 | — |
| diffusiongemma-26B-A4B-it google | author-reported Author reports MMMLU 81.5% | 25.8B (4B active) | 256K | — | Single high-end GPU | Apache 2.0 | 0 | — |
| gemma-4-31B google | author-reported Author reports MMMLU 88.4%decision-model estimate p=0.99 | 31.3B | 256K | $0.00 | Single high-end GPU | Apache 2.0 | 1 | — |
| Phi-4-mini-instruct microsoft | author-reported Author reports Multilingual MMLU (5-shot) 49.3Author reports MGSM (0-shot, CoT) 63.9 +1 | 3.84B | 128K | — | Single GPU | MIT | 0 | — |
| gemma-4-E4B google | author-reported Author reports MMMLU 76.6%decision-model estimate p=1.00 | 8.00B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| gemma-4-26B-A4B google | author-reported Author reports MMMLU 86.3%decision-model estimate p=0.99 | 25.8B (4B active) | 256K | $0.00 | Single high-end GPU | Apache 2.0 | 2 | — |
| SmolLM3-3B HuggingFaceTB | author-reported Author reports Global MMLU 64.1 | 3.08B | 64K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen2-72B-Instruct Qwen | author-reported Author reports C-Eval 83.8 | 72.7B | 32K | — | Multi-GPU node | tongyi-qianwen review | 2 | — |
| Apertus-8B-Instruct-2509 swiss-ai | author-reported Author reports XCOPA 69.8decision-model estimate p=1.00 +1 | 8.05B | 64K | $0.10 | Single GPU | Apache 2.0 | 2 | — |
| Qwen2-1.5B-Instruct Qwen | author-reported Author reports C-Eval 63.8 | 1.54B | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-235B-A22B-Thinking-2507 Qwen | author-reported Author reports MMLU-ProX 81.0Author reports INCLUDE 81.0 +1 | 235B (22B active) | 256K | $0.23 | Multi-GPU node | Apache 2.0 | 3 | — |
| gemma-4-E2B google | author-reported Author reports MMMLU 67.4%decision-model estimate p=0.99 | 5.12B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen2-0.5B-Instruct Qwen | author-reported Author reports C-Eval 45.2 | 494M | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen2-0.5B Qwen | author-reported Author reports C-Eval 58.2Author reports CMMLU 55.1 +1 | 494M | 128K | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-4-12B google | author-reported Author reports MMMLU 83.4%decision-model estimate p=0.99 | 12.0B | 256K | — | Single high-end GPU | Apache 2.0 | 0 | — |
| Qwen-7B-Chat Qwen | author-reported Author reports CEval (val) 59.7 | 7.72B | 32K | — | Single GPU | tongyi-qianwen-license-ag… review | 0 | — |
| gemma-4-31B-it-assistant google | author-reported Author reports MMMLU 88.4%decision-model estimate p=0.97 | 470M | 256K | — | Laptop / edge | Apache 2.0 | 0 | — |
| DeepSeek-V2-Lite deepseek-ai | author-reported Author reports C-Eval 60.3Author reports CMMLU 64.3 | 15.7B | 160K | — | Single high-end GPU | DeepSeek License (deepsee… attribution | 0 | — |
| gemma-3-4b-pt google | author-reported Author reports MGSM 34.7Author reports Global-MMLU-Lite 57.0 +1 | 4.30B | — | — | Single GPU | Gemma Terms of Use attribution | 1 | — |
| translategemma-4b-it google | author-reported Author reports WMT24++ (55 langs) MetricX ↓ 5.32decision-model estimate p=1.00 | 4.30B | 128K | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| Phi-3.5-MoE-instruct microsoft | author-reported Author reports Multilingual MMLU (5-shot) 69.9decision-model estimate p=0.93 +1 | 41.9B | 128K | — | Multi-GPU node | MIT | 0 | — |
| DeepSeek-V4-Flash-DSpark deepseek-ai | author-reported Author reports MMMLU (EM) 88.8Author reports C-Eval (EM) 92.1 +1 | 165B | 1M | — | Multi-GPU node | MIT | 0 | — |
| MiMo-V2.5-Pro XiaomiMiMo | author-reported Author reports C-Eval 91.5Author reports CMMLU 90.2 +1 | 1.02T | 1M | $0.30 | Cluster | MIT | 2 | — |
| Qwen2-7B Qwen | author-reported Author reports C-Eval 83.2Author reports CMMLU 83.9 +1 | 7.62B | 128K | — | Single GPU | Apache 2.0 | 1 | — |
| Qwen-72B Qwen | author-reported Author reports C-Eval 83.3Author reports CMMLU 83.6 | 72.3B | 32K | — | Multi-GPU node | tongyi-qianwen-license-ag… review | 0 | — |
| Qwen-7B Qwen | author-reported Author reports C-Eval 63.5 | 7.72B | 32K | — | Single GPU | tongyi-qianwen-license-ag… review | 0 | — |
| DeepSeek-V2-Lite-Chat deepseek-ai | author-reported Author reports C-Eval 60.3Author reports CMMLU 64.3 | 15.7B | 160K | — | Single high-end GPU | DeepSeek License (deepsee… attribution | 0 | — |
| gemma-4-26B-A4B-it-assistant google | author-reported Author reports MMMLU 86.3%decision-model estimate p=0.97 | 420M (4B active) | 256K | — | Laptop / edge | Apache 2.0 | 0 | — |
| gemma-3-1b-pt google | author-reported Author reports MGSM 2.04Author reports Global-MMLU-Lite 24.9 +1 | 1000M | 32K | — | Laptop / edge | Gemma Terms of Use attribution | 1 | — |
| Qwen2-1.5B Qwen | author-reported Author reports C-Eval 70.6Author reports CMMLU 70.3 +1 | 1.54B | 128K | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-3-12b-pt google | author-reported Author reports MGSM 64.3Author reports Global-MMLU-Lite 69.4 +1 | 12.2B | — | — | Single high-end GPU | Gemma Terms of Use attribution | 1 | — |
| DeepSeek-V2-Chat deepseek-ai | author-reported Author reports C-Eval 81.7Author reports CMMLU 84.0 | 236B | 160K | — | Multi-GPU node | DeepSeek License (deepsee… attribution | 0 | — |
| t5gemma-2-270m-270m google | author-reported Author reports MGSM 1.8Author reports Global-MMLU-Lite 23.5 +1 | 786M | — | — | Laptop / edge | Gemma Terms of Use attribution | 0 | — |
| LFM2-1.2B LiquidAI | author-reported Author reports MGSM 55.04Author reports MMMLU 46.73 | 1.17B | 125K | — | Laptop / edge | lfm1.0 review | 0 | — |
| Qwen2-72B Qwen | author-reported Author reports C-Eval 91.0Author reports CMMLU 90.1 | 72.7B | 128K | — | Multi-GPU node | tongyi-qianwen review | 1 | — |
| Moonlight-16B-A3B-Instruct moonshotai | author-reported Author reports C-Eval 77.2Author reports CMMLU 78.2 | 16.0B (3B active) | 8K | — | Single high-end GPU | MIT | 0 | — |
| translategemma-12b-it google | author-reported Author reports WMT24++ (55 langs) MetricX ↓ 5.32decision-model estimate p=1.00 | 12.2B | 128K | — | Single high-end GPU | Gemma Terms of Use attribution | 0 | — |
| internlm3-8b-instruct internlm | author-reported Author reports CMMLU(0-shot) 83.1 | 8.80B | 32K | — | Single GPU | Apache 2.0 | 0 | — |
| Hy3-preview tencent | author-reported Author reports C-Eval 89.80Author reports CMMLU 89.61 +1 | 299B | 256K | $0.18 | Cluster | Other review | 1 | — |
| DeepSeek-V2 deepseek-ai | author-reported Author reports C-Eval 81.7Author reports CMMLU 84.0 | 236B | 160K | — | Multi-GPU node | DeepSeek License (deepsee… attribution | 0 | — |
| pplx-decider-v1-27b perplexity-ai | author-reported Author reports Belebele 94.00% | 26.1B | 256K | — | Single high-end GPU | Apache 2.0 | 0 | — |
| bge-m3 BAAI | structural decision-model estimate p=1.00name: m3 | — | 8K | $0.01 | — | MIT | 3 | — |
| Qwen3-VL-8B-Instruct Qwen | structural decision-model estimate p=0.96 | 8.77B | 256K | $0.25 | Single GPU | Apache 2.0 | 4 | — |
| Qwen3-8B Qwen | structural decision-model estimate p=1.00 | 8.19B | 40K | $0.07 | Single GPU | Apache 2.0 | 3 | — |
| Qwen2.5-7B-Instruct Qwen | structural decision-model estimate p=0.96 | 7.62B | 32K | $0.10 | Single GPU | Apache 2.0 | 3 | — |
| MiniMax-M3 MiniMaxAI | structural name: M3 | 427B | 1M | $0.22 | Cluster | MiniMax Model License (mi… gated-threshold | 7 | — |
| Qwen3-14B Qwen | structural decision-model estimate p=1.00 | 14.8B | 40K | $0.07 | Single high-end GPU | Apache 2.0 | 4 | — |
| gemma-3-4b-it google | structural decision-model estimate p=1.00 | 4.30B | 128K | $0.04 | Single GPU | Gemma Terms of Use attribution | 4 | — |
| Qwen3-0.6B Qwen | structural decision-model estimate p=0.99 | 752M | 40K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-32B Qwen | structural decision-model estimate p=1.00 | 32.8B | 40K | $0.08 | Single high-end GPU | Apache 2.0 | 4 | — |
| paraphrase-multilingual-MiniLM-L12-v2 sentence-transformers | structural decision-model estimate p=1.00name: multilingual +1 | 118M | 512 | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-3-27b-it google | structural decision-model estimate p=1.00 | 27.4B | 128K | $0.08 | Single high-end GPU | Gemma Terms of Use attribution | 4 | — |
| multilingual-e5-large intfloat | structural decision-model estimate p=1.00name: multilingual +1 | 560M | 514 | $0.01 | Laptop / edge | MIT | 2 | — |
| embeddinggemma-300m google | structural decision-model estimate p=1.00 | 303M | 2K | $0.0020 | Laptop / edge | Gemma Terms of Use attribution | 2 | — |
| Qwen3-235B-A22B Qwen | structural decision-model estimate p=1.00 | 235B (22B active) | 40K | $0.20 | Multi-GPU node | Apache 2.0 | 4 | — |
| Qwen2.5-1.5B-Instruct Qwen | structural decision-model estimate p=0.94 | 1.54B | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-VL-30B-A3B-Instruct Qwen | structural decision-model estimate p=0.98 | 31.1B (3B active) | 256K | $0.15 | Single high-end GPU | Apache 2.0 | 4 | — |
| Qwen3-Embedding-0.6B Qwen | structural decision-model estimate p=1.00 | 596M | 32K | $0.01 | Laptop / edge | Apache 2.0 | 1 | — |
| DeepSeek-OCR deepseek-ai | structural tag: multilingual | 3.34B | 8K | — | Single GPU | MIT | 1 | — |
| Llama-4-Scout-17B-16E-Instruct meta-llama | structural decision-model estimate p=0.7412 languages | 109B | 1.3M | $0.08 | Multi-GPU node | Llama Community License (… attribution | 4 | — |
| Qwen2.5-3B-Instruct Qwen | structural decision-model estimate p=0.92 | 3.09B | 32K | — | Single GPU | Qwen License (qwen-resear… gated-threshold | 1 | — |
| Qwen2.5-0.5B-Instruct Qwen | structural decision-model estimate p=0.92 | 494M | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-3-12b-it google | structural decision-model estimate p=1.00 | 12.2B | 128K | $0.04 | Single high-end GPU | Gemma Terms of Use attribution | 3 | — |
| Qwen3-4B Qwen | structural decision-model estimate p=0.99 | 4.02B | 40K | — | Single GPU | Apache 2.0 | 1 | — |
| Qwen3-Embedding-8B Qwen | structural decision-model estimate p=1.00 | 7.57B | 40K | $0.01 | Single GPU | Apache 2.0 | 2 | — |
| Qwen3-4B-Instruct-2507 Qwen | structural decision-model estimate p=0.75 | 4.02B | 256K | $0.01 | Single GPU | Apache 2.0 | 1 | — |
| Unlimited-OCR baidu | structural decision-model estimate p=0.73tag: multilingual | 3.34B | 32K | — | Single GPU | MIT | 0 | — |
| gemma-3-1b-it google | structural decision-model estimate p=0.99 | 1000M | 32K | — | Laptop / edge | Gemma Terms of Use attribution | 1 | — |
| Qwen3-VL-4B-Instruct Qwen | structural decision-model estimate p=0.96 | 4.44B | 256K | — | Single GPU | Apache 2.0 | 2 | — |
| Qwen3-30B-A3B Qwen | structural decision-model estimate p=1.00 | 30.5B (3B active) | 40K | $0.12 | Single high-end GPU | Apache 2.0 | 2 | — |
| Qwen3-VL-2B-Instruct Qwen | structural decision-model estimate p=0.93 | 2.13B | 256K | — | Single GPU | Apache 2.0 | 1 | — |
| Qwen3-235B-A22B-Instruct-2507 Qwen | structural decision-model estimate p=0.80 | 235B (22B active) | 256K | $0.09 | Multi-GPU node | Apache 2.0 | 5 | — |
| Qwen2.5-72B-Instruct Qwen | structural decision-model estimate p=0.98 | 72.7B | 32K | $0.36 | Multi-GPU node | Qwen License (qwen) gated-threshold | 3 | — |
| multilingual-e5-small intfloat | structural decision-model estimate p=1.00name: multilingual +1 | 118M | 512 | — | Laptop / edge | MIT | 1 | — |
| Qwen3-1.7B Qwen | structural decision-model estimate p=0.99 | 2.03B | 40K | — | Single GPU | Apache 2.0 | 1 | — |
| Qwen3-30B-A3B-Instruct-2507 Qwen | structural decision-model estimate p=0.80 | 30.5B (3B active) | 256K | $0.09 | Single high-end GPU | Apache 2.0 | 2 | — |
| Qwen3-Reranker-0.6B Qwen | structural decision-model estimate p=1.00 | 596M | 40K | $0.01 | Laptop / edge | Apache 2.0 | 2 | — |
| multilingual-e5-large-instruct intfloat | structural decision-model estimate p=1.00name: multilingual +1 | 560M | 514 | $0.01 | Laptop / edge | MIT | 2 | — |
| paraphrase-multilingual-mpnet-base-v2 sentence-transformers | structural decision-model estimate p=1.00name: multilingual +1 | 278M | 514 | — | Laptop / edge | Apache 2.0 | 1 | — |
| Muse-Glimmer-30B meta-models | structural decision-model estimate p=1.00 | 29.8B | 128K | $0.30 | Single high-end GPU | Apache 2.0 | 4 | — |
| Qwen2.5-0.5B Qwen | structural decision-model estimate p=0.88 | 494M | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-VL-235B-A22B-Instruct Qwen | structural decision-model estimate p=0.98 | 236B (22B active) | 256K | $0.20 | Multi-GPU node | Apache 2.0 | 3 | — |
| multilingual-e5-base intfloat | structural decision-model estimate p=1.00name: multilingual +1 | 278M | 514 | — | Laptop / edge | MIT | 1 | — |
| Qwen3-VL-Embedding-2B Qwen | structural decision-model estimate p=1.00 | 2.13B | 256K | — | Single GPU | Apache 2.0 | 1 | — |
| jina-embeddings-v3 jinaai | structural decision-model estimate p=1.00tag: multilingual +1 | 572M | 8K | — | Laptop / edge | CC BY-NC 4.0 non-commercial | 0 | — |
| Qwen3-Embedding-4B Qwen | structural decision-model estimate p=1.00 | 4.02B | 40K | $0.02 | Single GPU | Apache 2.0 | 1 | — |
| Qwen3-Reranker-4B Qwen | structural decision-model estimate p=1.00 | 4.02B | 40K | $0.03 | Single GPU | Apache 2.0 | 1 | — |
| Qwen2.5-14B-Instruct Qwen | structural decision-model estimate p=0.97 | 14.8B | 32K | — | Single high-end GPU | Apache 2.0 | 1 | — |
| Qwen2.5-32B-Instruct Qwen | structural decision-model estimate p=0.93 | 32.8B | 32K | — | Single high-end GPU | Apache 2.0 | 1 | — |
| DeepSeek-OCR-2 deepseek-ai | structural decision-model estimate p=0.71tag: multilingual | 3.39B | 8K | — | Single GPU | Apache 2.0 | 0 | — |
| distiluse-base-multilingual-cased-v2 sentence-transformers | structural decision-model estimate p=1.00name: multilingual +1 | 135M | 512 | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-VL-32B-Instruct Qwen | structural decision-model estimate p=0.95 | 33.4B | 256K | — | Multi-GPU node | Apache 2.0 | 2 | — |
| snowflake-arctic-embed-l-v2.0 Snowflake | structural decision-model estimate p=1.0074 languages | 568M | 8K | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-VL-Embedding-8B Qwen | structural decision-model estimate p=1.00 | 8.14B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen2.5-7B Qwen | structural decision-model estimate p=0.94 | 7.62B | 128K | $0.10 | Single GPU | Apache 2.0 | 1 | — |
| distiluse-base-multilingual-cased-v1 sentence-transformers | structural decision-model estimate p=1.00name: multilingual +1 | 135M | 512 | — | Laptop / edge | Apache 2.0 | 1 | — |
| Qwen3-VL-235B-A22B-Thinking Qwen | structural decision-model estimate p=0.93 | 236B (22B active) | 256K | — | Multi-GPU node | Apache 2.0 | 3 | — |
| Qwen3-Omni-30B-A3B-Instruct Qwen | structural decision-model estimate p=1.00 | 35.3B (3B active) | — | — | Multi-GPU node | Apache 2.0 (apache-2.0) | 0 | — |
| Qwen2.5-1.5B Qwen | structural decision-model estimate p=0.91 | 1.54B | 128K | — | Laptop / edge | Apache 2.0 | 1 | — |
| gemma-3-270m google | structural decision-model estimate p=0.99 | 268M | 32K | — | Laptop / edge | Gemma Terms of Use attribution | 0 | — |
| Phi-3.5-mini-instruct microsoft | structural decision-model estimate p=0.88tag: multilingual | 3.82B | 128K | — | Single GPU | MIT | 0 | — |
| HunyuanOCR tencent | structural tag: multilingual | 1.12B | 128K | — | Laptop / edge | tencent-hunyuan-community review | 0 | — |
| Qwen3-0.6B-Base Qwen | structural decision-model estimate p=0.97 | 596M | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| nomic-embed-text-v2-moe nomic-ai | structural decision-model estimate p=1.00101 languages | 475M | 2K | — | Laptop / edge | Apache 2.0 | 0 | — |
| Phi-3.5-vision-instruct microsoft | structural tag: multilingual | 4.15B | 128K | — | Single GPU | MIT | 0 | — |
| gte-multilingual-base Alibaba-NLP | structural decision-model estimate p=1.00name: multilingual +1 | 305M | 8K | — | Laptop / edge | Apache 2.0 | 0 | — |
| Qwen3-4B-Base Qwen | structural decision-model estimate p=0.99 | 4.02B | 32K | — | Single GPU | Apache 2.0 | 1 | — |
| GOT-OCR2_0 stepfun-ai | structural decision-model estimate p=0.77tag: multilingual | 716M | 32K | — | Laptop / edge | Apache 2.0 | 0 | — |
| Qwen3-Reranker-8B Qwen | structural decision-model estimate p=1.00 | 8.19B | 40K | $0.05 | Single GPU | Apache 2.0 | 1 | — |
| embeddinggemma-2 google | structural tag: multilingual | 744M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| Qwen3-8B-Base Qwen | structural decision-model estimate p=0.98 | 8.19B | 32K | — | Single GPU | Apache 2.0 | 1 | — |
| Qwen3-VL-8B-Thinking Qwen | structural decision-model estimate p=0.93 | 8.77B | 256K | — | Single GPU | Apache 2.0 | 3 | — |
| bloom bigscience | structural decision-model estimate p=1.0046 languages | 176B | — | — | Multi-GPU node | BLOOM RAIL 1.0 attribution | 0 | — |
| clip-ViT-B-32-multilingual-v1 sentence-transformers | structural decision-model estimate p=1.00name: multilingual +1 | 135M | 512 | $0.0050 | Laptop / edge | Apache 2.0 | 2 | — |
| jina-reranker-v2-base-multilingual jinaai | structural decision-model estimate p=0.99name: multilingual +1 | 278M | 1K | — | Laptop / edge | CC BY-NC 4.0 non-commercial | 0 | — |
| Qwen2.5-3B Qwen | structural decision-model estimate p=0.92 | 3.09B | 32K | — | Single GPU | Qwen License (qwen-resear… gated-threshold | 1 | — |
| LaBSE sentence-transformers | structural decision-model estimate p=1.00tag: multilingual +1 | 471M | 512 | — | Laptop / edge | Apache 2.0 | 1 | — |
| LightOnOCR-2-1B lightonai | structural decision-model estimate p=1.0011 languages | 1.01B | 16K | — | Laptop / edge | Apache 2.0 | 1 | — |
| w2v-bert-2.0 facebook | structural decision-model estimate p=1.0096 languages | 580M | — | — | Laptop / edge | MIT | 0 | — |
| paligemma-3b-pt-224 google | structural decision-model estimate p=0.89 | 2.92B | — | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| Qwen3-1.7B-Base Qwen | structural decision-model estimate p=0.97 | 1.72B | 32K | — | Single GPU | Apache 2.0 | 1 | — |
| bloomz-560m bigscience | structural decision-model estimate p=0.9946 languages | 559M | 2K | — | Laptop / edge | BLOOM RAIL 1.0 attribution | 0 | — |
| bge-multilingual-gemma2 BAAI | structural decision-model estimate p=0.99name: multilingual | 9.24B | 8K | — | Single GPU | Gemma Terms of Use attribution | 1 | — |
| gemma-3-270m-it google | structural decision-model estimate p=1.00 | 268M | 32K | — | Laptop / edge | Gemma Terms of Use attribution | 0 | — |
| canine-c google | structural decision-model estimate p=0.98tag: multilingual +1 | 132M | — | — | Laptop / edge | Apache 2.0 | 1 | — |
| LFM2.5-2.6B LiquidAI | structural decision-model estimate p=1.0016 languages | 2.70B | 128K | $0.00 | Single GPU | lfm1.0 review | 1 | — |
| Qwen3Guard-Gen-0.6B Qwen | structural decision-model estimate p=1.00 | 752M | 32K | — | Laptop / edge | Apache 2.0 | 1 | — |
| chandra-ocr-2 datalab-to | structural decision-model estimate p=1.00 | 5.30B | 256K | — | Single GPU | OpenRAIL attribution | 0 | — |
| bloom-560m bigscience | structural decision-model estimate p=1.0048 languages | 559M | — | — | Laptop / edge | BLOOM RAIL 1.0 attribution | 0 | — |
| Qwen2.5-14B Qwen | structural decision-model estimate p=0.95 | 14.8B | 128K | — | Single high-end GPU | Apache 2.0 | 1 | — |
| Qianfan-OCR baidu | structural decision-model estimate p=1.00tag: multilingual | 4.74B | 32K | — | Single GPU | Apache 2.0 | 0 | — |
| Llama-4-Maverick-17B-128E-Instruct meta-llama | structural decision-model estimate p=0.7912 languages | 402B | 512K | $0.19 | Cluster | Llama Community License (… attribution | 2 | — |
| Qwen3-VL-4B-Thinking Qwen | structural decision-model estimate p=0.94 | 4.44B | 256K | — | Single GPU | Apache 2.0 | 2 | — |
| Qwen2.5-32B Qwen | structural decision-model estimate p=0.96 | 32.8B | 128K | — | Single high-end GPU | Apache 2.0 | 1 | — |
| surya-ocr-2 datalab-to | structural decision-model estimate p=1.00 | 686M | 256K | — | Laptop / edge | OpenRAIL attribution | 1 | — |
| Qwen3-VL-30B-A3B-Thinking Qwen | structural decision-model estimate p=0.93 | 31.1B (3B active) | 256K | $0.29 | Single high-end GPU | Apache 2.0 | 2 | — |
| gemma-3n-E2B-it google | structural decision-model estimate p=0.98 | 5.44B | 32K | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| LFM2.5-8B-A1B LiquidAI | structural decision-model estimate p=1.0010 languages | 8.47B (1B active) | 125K | — | Single GPU | lfm1.0 review | 0 | — |
| Qwen3Guard-Gen-8B Qwen | structural decision-model estimate p=1.00 | 8.19B | 32K | — | Single GPU | Apache 2.0 | 1 | — |
| gemma-3n-E4B-it google | structural decision-model estimate p=1.00 | 7.85B | 32K | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| LFM2.5-24B-A2B LiquidAI | structural decision-model estimate p=0.7416 languages | 24.0B (2B active) | — | — | Single high-end GPU | lfm1.0 review | 0 | — |
| Qwen3-14B-Base Qwen | structural decision-model estimate p=0.98 | 14.8B | 32K | — | Single high-end GPU | Apache 2.0 | 1 | — |
| Phi-3-vision-128k-instruct microsoft | structural tag: multilingual | 4.15B | 128K | — | Single GPU | MIT | 0 | — |
| MiniCPM-o-2_6 openbmb | structural tag: multilingual | 8.67B | 32K | — | Single GPU | Apache 2.0 | 0 | — |
| dots.ocr dots-studio | structural tag: multilingual | 3.04B | 128K | — | Single GPU | MIT | 0 | — |
| Qwen3.5-0.8B-Base Qwen | structural decision-model estimate p=0.94 | 873M | 256K | — | Laptop / edge | Apache 2.0 | 0 | — |
| granite-4.1-8b ibm-granite | structural decision-model estimate p=0.99 | 8.79B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen3-Omni-30B-A3B-Thinking Qwen | structural decision-model estimate p=1.00 | 31.7B (3B active) | — | — | Single high-end GPU | Apache 2.0 (apache-2.0) | 0 | — |
| granite-4.1-3b ibm-granite | structural decision-model estimate p=0.99 | 3.40B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen3.5-4B-Base Qwen | structural decision-model estimate p=0.95 | 4.66B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
| paligemma-3b-mix-224 google | structural decision-model estimate p=0.82 | 2.92B | — | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| d1-3B LiquidAI | structural decision-model estimate p=0.7616 languages | 3.12B | — | — | Single GPU | lfm1.0 review | 0 | — |
| Qwen3.5-9B-Base Qwen | structural decision-model estimate p=0.97 | 9.65B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
| Qwen3.5-2B-Base Qwen | structural decision-model estimate p=0.92 | 2.27B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
| ReaderLM-v2 jinaai | structural decision-model estimate p=1.00tag: multilingual | 1.54B | 501K | — | Laptop / edge | CC BY-NC 4.0 non-commercial | 1 | — |
| MiniCPM-V-4_5 openbmb | structural decision-model estimate p=1.00tag: multilingual | 8.70B | 40K | — | Single GPU | Apache 2.0 | 0 | — |
| paligemma2-3b-pt-224 google | structural decision-model estimate p=0.71 | 3.03B | — | — | Single GPU | Gemma Terms of Use attribution | 0 | — |
| InternVL3-78B OpenGVLab | structural decision-model estimate p=0.71tag: multilingual | 78.4B | — | — | Multi-GPU node | Qwen License (qwen) gated-threshold | 0 | — |
| granite-4.2-8b ibm-granite | structural decision-model estimate p=1.0012 languages | 8.79B | 128K | $0.06 | Single GPU | Apache 2.0 | 2 | — |
| aya-expanse-8b CohereLabs | structural decision-model estimate p=0.90name: aya +1 | 8.03B | — | — | Single GPU | CC BY-NC 4.0 non-commercial | 0 | — |
| LFM2.5-230M LiquidAI | structural decision-model estimate p=1.0010 languages | 230M | 125K | — | Laptop / edge | lfm1.0 review | 0 | — |
| Qwen3Guard-Gen-4B Qwen | structural decision-model estimate p=1.00 | 4.41B | 32K | — | Single GPU | Apache 2.0 | 1 | — |
| granite-4.0-h-tiny ibm-granite | structural decision-model estimate p=0.99 | 6.94B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| Bielik-11B-v3.0-Instruct speakleash | structural decision-model estimate p=0.98tag: multilingual +1 | 11.2B | — | $0.40 | Single high-end GPU | Apache 2.0 | 1 | — |
| Qwen3-VL-2B-Thinking Qwen | structural decision-model estimate p=0.93 | 2.13B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
| c4ai-command-r7b-12-2024 CohereLabs | structural decision-model estimate p=0.8323 languages | 8.03B | — | — | Single GPU | CC BY-NC 4.0 non-commercial | 1 | — |
| MiniCPM-Llama3-V-2_5 openbmb | structural decision-model estimate p=1.00tag: multilingual | 8.54B | 8K | — | Single GPU | Not declared review | 1 | — |
| granite-4.0-h-small ibm-granite | structural decision-model estimate p=0.99 | 32.2B | 128K | — | Single high-end GPU | Apache 2.0 | 0 | — |
| InternVL3-8B OpenGVLab | structural tag: multilingual | 7.94B | — | — | Single GPU | Apache 2.0 (qwen) | 0 | — |
| granite-3.3-8b-instruct ibm-granite | structural decision-model estimate p=0.99 | 8.17B | 128K | — | Single GPU | Apache 2.0 | 0 | — |
| InternVL3-1B OpenGVLab | structural tag: multilingual | 938M | — | — | Laptop / edge | Apache 2.0 (qwen) | 0 | — |
| PaddleOCR-VL-1.6 PaddlePaddle | structural tag: multilingual | 959M | 128K | — | Laptop / edge | Apache 2.0 | 0 | — |
| c4ai-command-r-v01 CohereLabs | structural decision-model estimate p=0.7910 languages | 35.0B | — | — | Multi-GPU node | CC BY-NC 4.0 non-commercial | 0 | — |
| Phi-3-medium-4k-instruct microsoft | structural tag: multilingual | 14.0B | — | — | Single high-end GPU | MIT | 1 | — |
| GOT-OCR-2.0-hf stepfun-ai | structural decision-model estimate p=0.82tag: multilingual | 561M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| laya convaiinnovations | structural decision-model estimate p=1.00 | 421M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| laya-multilingual convaiinnovations | structural decision-model estimate p=1.00name: multilingual +1 | 322M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| gliner2.5-multi-v1 fastino | structural decision-model estimate p=0.96tag: multilingual | 287M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| Julia-1 SupersonicLabs | structural decision-model estimate p=0.93tag: multilingual | 144M | — | — | Laptop / edge | Apache 2.0 | 0 | — |
| LFM2.5-2.6B-RLCD monotykamary | structural decision-model estimate p=0.7316 languages | 2.70B | 128K | — | Single GPU | lfm1.0 review | 0 | — |
| metask-jev-4b-policy-mix wayfind | structural decision-model estimate p=1.00tag: multilingual +1 | 4.54B | 256K | — | Single GPU | Apache 2.0 | 0 | — |
No model matches every constraint. Loosen one, or switch the evidence filter back to any.
How to read this
- measured a third-party leaderboard row for this model. Settings (agent harness, edit format, prompt) differ between boards; compare within a benchmark, not across.
- author-reported a number the model author published in the card or eval files. Not reproduced by us.
- structural a deterministic signal from the name, config or chat template, or a decision-model estimate. A weak signal: it says the model was built or tagged for the job, not how well it does it. "decision-model estimate" chips are a probability from the Jev decision model, not a measurement.
- Runs on is the memory class of the BF16 weights (method); choosing a class shows that class and smaller. Quantized builds may fit a class lower.
- Price from is the cheapest observed hosted input price, USD per million tokens; models with no observed price are hidden when a ceiling is set. License friction: permissive = none or attribution-only; high = non-commercial.
- Ordering within a tier is the watch-list score (Hub likes, downloads, recency), not a quality verdict.
Subscribe
Follow "Work across languages"
- /feeds/for/multilingual.xml(RSS 2.0) · one item per model with measured or author-reported evidence for this job
- /feeds/for/multilingual.json(JSON Feed 1.1) · same items, JSON Feed
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.