{"total":29,"models":[{"id":"gemini-4-argon-high","name":"Gemini 4 Argon (High)","creator":"Google DeepMind","creatorSlug":"google","licenseType":"proprietary","category":"reasoning","arenaElo":1532,"arenaCodingElo":1678,"arenaHardElo":1542,"arenaAgentElo":1530,"voteCount":4900,"intelligenceIndex":99.2,"sweBenchScore":82.4,"gpqaScore":86.5,"mmluProScore":93.8,"livebenchScore":88.2,"scaleSealScore":99,"ifEvalScore":95.6,"outputSpeedTps":115,"latencyTtftMs":380,"qualitySpeedRatio":112,"contextWindowTokens":2000000,"architectureType":"Hybrid Reasoning","parameters":"Frontier Multimodal","inputPricePerM":2.5,"outputPricePerM":10,"blendedPricePerM":4.38,"cacheReadPricePerM":0.25,"releaseDate":"2026-09","description":"Number 1 overall leader worldwide on LMSYS Arena.ai. Frontier native reasoning with 2M token context, deep multimodal understanding, and unmatched recursive problem-solving.","skillsTopic":"gemini","docsUrl":"https://yakaai.com/doc/google-gemini/gemini-api","rankComposite":1},{"id":"claude-opus-5-5-high","name":"Claude Opus 5.5 (High)","creator":"Anthropic","creatorSlug":"anthropic","licenseType":"proprietary","category":"reasoning","arenaElo":1516,"arenaCodingElo":1813,"arenaHardElo":1525,"arenaAgentElo":1545,"voteCount":9496,"intelligenceIndex":99,"sweBenchScore":81.8,"gpqaScore":85.4,"mmluProScore":93.2,"livebenchScore":87.5,"scaleSealScore":98.8,"ifEvalScore":95.2,"outputSpeedTps":88,"latencyTtftMs":420,"qualitySpeedRatio":89,"contextWindowTokens":500000,"architectureType":"Hybrid Reasoning","parameters":"Ultra-MoE","inputPricePerM":5,"outputPricePerM":25,"blendedPricePerM":10,"cacheReadPricePerM":0.5,"releaseDate":"2026-09","description":"Rank #2 overall worldwide and #1 Coding Champion (1813 Elo) and #1 Autonomous Agent on Arena. Unrivaled repository refactoring and autonomous tool chaining.","skillsTopic":"claude","docsUrl":"https://yakaai.com/doc/anthropics/anthropic-sdk-typescript","rankComposite":2},{"id":"claude-fable-5-1-max","name":"Claude Fable 5.1 (Max)","creator":"Anthropic","creatorSlug":"anthropic","licenseType":"proprietary","category":"reasoning","arenaElo":1512,"arenaCodingElo":1743,"arenaHardElo":1518,"arenaAgentElo":1520,"voteCount":15623,"intelligenceIndex":98.6,"sweBenchScore":79.5,"gpqaScore":84.8,"mmluProScore":92.8,"livebenchScore":86.8,"scaleSealScore":98.4,"ifEvalScore":94.9,"outputSpeedTps":92,"latencyTtftMs":440,"qualitySpeedRatio":91,"contextWindowTokens":500000,"architectureType":"Hybrid Reasoning","parameters":"Extended MoE","inputPricePerM":4,"outputPricePerM":20,"blendedPricePerM":8,"cacheReadPricePerM":0.4,"releaseDate":"2026-08","description":"Rank #3 overall worldwide on LMSYS Arena.ai. Specialized in extended deliberation, deep creative prose, and architectural systems synthesis.","skillsTopic":"claude","docsUrl":"https://yakaai.com/doc/anthropics/anthropic-sdk-typescript","rankComposite":3},{"id":"claude-opus-5-high","name":"Claude Opus 5 (High)","creator":"Anthropic","creatorSlug":"anthropic","licenseType":"proprietary","category":"reasoning","arenaElo":1503,"arenaCodingElo":1659,"arenaHardElo":1512,"arenaAgentElo":1535,"voteCount":69307,"intelligenceIndex":98.4,"sweBenchScore":78.6,"gpqaScore":84.2,"mmluProScore":92.5,"livebenchScore":86.4,"scaleSealScore":98.2,"ifEvalScore":94.8,"outputSpeedTps":84,"latencyTtftMs":460,"qualitySpeedRatio":83,"contextWindowTokens":500000,"architectureType":"Hybrid Reasoning","parameters":"Ultra-MoE","inputPricePerM":5,"outputPricePerM":25,"blendedPricePerM":10,"cacheReadPricePerM":0.5,"releaseDate":"2026-08","description":"Number 1 overall leader on Arena.ai and Agent benchmarks. Features deep recursive thinking, dynamic search integration, and unmatched complex repository refactoring.","skillsTopic":"claude","docsUrl":"https://yakaai.com/doc/anthropics/anthropic-sdk-typescript","rankComposite":4},{"id":"gemini-3-8-flash-high","name":"Gemini 3.8 Flash (High)","creator":"Google DeepMind","creatorSlug":"google","licenseType":"proprietary","category":"llm","arenaElo":1499,"arenaCodingElo":1585,"arenaHardElo":1498,"arenaAgentElo":1495,"voteCount":41423,"intelligenceIndex":97.8,"sweBenchScore":76.2,"gpqaScore":81.5,"mmluProScore":89.4,"livebenchScore":84.1,"scaleSealScore":96.8,"ifEvalScore":93.4,"outputSpeedTps":185,"latencyTtftMs":210,"qualitySpeedRatio":162,"contextWindowTokens":2000000,"architectureType":"Dense","parameters":"High-Efficiency Frontier","inputPricePerM":0.15,"outputPricePerM":0.6,"blendedPricePerM":0.26,"cacheReadPricePerM":0.04,"releaseDate":"2026-08","description":"Ultra-fast frontier model with unmatched quality-to-speed ratio (162 pts) and 2M token context window at $0.26/M blended price.","skillsTopic":"gemini","docsUrl":"https://yakaai.com/doc/google-gemini/gemini-api","rankComposite":5},{"id":"mimo-v2-6-pro","name":"MiMo-v2.6 Pro","creator":"Xiaomi","creatorSlug":"xiaomi","licenseType":"proprietary","category":"reasoning","arenaElo":1491,"arenaCodingElo":1625,"arenaHardElo":1490,"arenaAgentElo":1485,"voteCount":4065,"intelligenceIndex":97.2,"sweBenchScore":74.5,"gpqaScore":79.8,"mmluProScore":88.5,"livebenchScore":83.2,"scaleSealScore":95.8,"ifEvalScore":92.1,"outputSpeedTps":105,"latencyTtftMs":350,"qualitySpeedRatio":98,"contextWindowTokens":256000,"architectureType":"Hybrid Reasoning","parameters":"High-Throughput MoE","inputPricePerM":0.8,"outputPricePerM":3.2,"blendedPricePerM":1.4,"cacheReadPricePerM":0.15,"releaseDate":"2026-08","description":"Top 10 frontier reasoning model from Xiaomi featuring aggressive test-time compute and agile mobile-edge agent execution.","skillsTopic":"general","docsUrl":"https://yakaai.com/leaderboard","rankComposite":6},{"id":"gemini-3-7-flash-high","name":"Gemini 3.7 Flash (High)","creator":"Google DeepMind","creatorSlug":"google","licenseType":"proprietary","category":"reasoning","arenaElo":1489,"arenaCodingElo":1591,"arenaHardElo":1488,"arenaAgentElo":1492,"voteCount":29177,"intelligenceIndex":95.9,"sweBenchScore":73.5,"gpqaScore":80.4,"mmluProScore":89.6,"livebenchScore":82.8,"scaleSealScore":95,"ifEvalScore":92.5,"outputSpeedTps":165,"latencyTtftMs":190,"qualitySpeedRatio":158,"contextWindowTokens":2000000,"architectureType":"MoE","parameters":"Fast-MoE","inputPricePerM":0.35,"outputPricePerM":1.4,"blendedPricePerM":0.61,"cacheReadPricePerM":0.08,"releaseDate":"2026-08","description":"Fastest reasoning model on the market delivering 165 tps throughput, 2M token context, and sub-200ms latency at ultra-competitive pricing.","skillsTopic":"gemini","docsUrl":"https://yakaai.com/doc/google-gemini/generative-ai-js","rankComposite":7},{"id":"qwen3-8-max","name":"Qwen 3.8 Max","creator":"Alibaba Qwen","creatorSlug":"qwen","licenseType":"proprietary","category":"llm","arenaElo":1483,"arenaCodingElo":1672,"arenaHardElo":1478,"voteCount":29418,"intelligenceIndex":93.6,"sweBenchScore":70.4,"gpqaScore":77.8,"mmluProScore":87.2,"livebenchScore":79.4,"scaleSealScore":92.8,"ifEvalScore":92,"outputSpeedTps":115,"latencyTtftMs":260,"qualitySpeedRatio":108,"contextWindowTokens":256000,"architectureType":"MoE","parameters":"Mega-MoE","inputPricePerM":0.6,"outputPricePerM":2.4,"blendedPricePerM":1.05,"releaseDate":"2026-08","description":"Alibaba flagship commercial model leading international multilingual benchmarks across English, Chinese, French, Arabic, and Japanese.","skillsTopic":"qwen","docsUrl":"https://yakaai.com/doc/QwenLM/Qwen2.5","rankComposite":8},{"id":"muse-spark-1-2-xhigh","name":"Muse-Spark 1.2 (xHigh)","creator":"Meta","creatorSlug":"meta","licenseType":"proprietary","category":"reasoning","arenaElo":1482,"arenaCodingElo":1534,"arenaHardElo":1484,"voteCount":9578,"intelligenceIndex":94.8,"sweBenchScore":72.1,"gpqaScore":79.5,"mmluProScore":88.7,"livebenchScore":81.2,"scaleSealScore":94.1,"ifEvalScore":91.8,"outputSpeedTps":125,"latencyTtftMs":240,"qualitySpeedRatio":119,"contextWindowTokens":500000,"architectureType":"MoE","parameters":"Spark-MoE","inputPricePerM":1,"outputPricePerM":4,"blendedPricePerM":1.75,"releaseDate":"2026-08","description":"Meta next-generation commercial reasoning engine trained with multimodal perception and verifiable symbolic mathematics.","rankComposite":9},{"id":"grok-4-6-thinking","name":"Grok 4.6 (Thinking)","creator":"xAI","creatorSlug":"xai","licenseType":"proprietary","category":"reasoning","arenaElo":1480,"arenaCodingElo":1482,"arenaHardElo":1475,"voteCount":14200,"intelligenceIndex":93.1,"sweBenchScore":69.5,"gpqaScore":77.1,"mmluProScore":86.5,"livebenchScore":78.6,"scaleSealScore":92,"ifEvalScore":90.8,"outputSpeedTps":92,"latencyTtftMs":340,"qualitySpeedRatio":86,"contextWindowTokens":256000,"architectureType":"Hybrid Reasoning","parameters":"Grok-MoE","inputPricePerM":2,"outputPricePerM":10,"blendedPricePerM":4,"releaseDate":"2026-08","description":"xAI flagship model integrated with Colossus compute cluster, featuring real-time verified X platform search and mathematical proof generation.","rankComposite":10},{"id":"kimi-k3-max","name":"Kimi K3 (Max)","creator":"Moonshot AI","creatorSlug":"moonshot","licenseType":"proprietary","category":"agent","arenaElo":1476,"arenaCodingElo":1655,"arenaHardElo":1482,"arenaAgentElo":1518,"voteCount":34325,"intelligenceIndex":94.2,"sweBenchScore":71.8,"gpqaScore":78.9,"mmluProScore":88,"livebenchScore":80.5,"scaleSealScore":93.5,"ifEvalScore":93.1,"outputSpeedTps":105,"latencyTtftMs":290,"qualitySpeedRatio":99,"contextWindowTokens":1000000,"architectureType":"MoE","parameters":"1M-MoE","inputPricePerM":0.8,"outputPricePerM":3.2,"blendedPricePerM":1.4,"releaseDate":"2026-07","description":"Long-context agentic reasoning champion with 1M tokens needle-in-a-haystack accuracy and deep multi-hop web research capabilities.","rankComposite":11},{"id":"glm-5-3-max","name":"GLM 5.3 Max (MIT)","creator":"Zhipu AI","creatorSlug":"zhipu","licenseType":"open-source","category":"llm","arenaElo":1474,"arenaCodingElo":1621,"arenaHardElo":1470,"voteCount":24035,"intelligenceIndex":92.4,"sweBenchScore":68.2,"gpqaScore":76.2,"mmluProScore":85.8,"livebenchScore":77.8,"scaleSealScore":91.2,"ifEvalScore":91.5,"outputSpeedTps":88,"latencyTtftMs":360,"qualitySpeedRatio":81,"contextWindowTokens":256000,"architectureType":"MoE","parameters":"380B (42B active)","minVramGb":48,"inputPricePerM":0.4,"outputPricePerM":1.2,"blendedPricePerM":0.6,"releaseDate":"2026-08","description":"Top-ranked permissive MIT open-source model matching commercial frontier quality on both coding and bilingual reasoning benchmarks.","hfModelId":"THUDM/glm-5.3-max","rankComposite":12},{"id":"claude-sonnet-5-5-xhigh","name":"Claude Sonnet 5.5 (xHigh)","creator":"Anthropic","creatorSlug":"anthropic","licenseType":"proprietary","category":"coding","arenaElo":1473,"arenaCodingElo":1768,"arenaHardElo":1485,"arenaAgentElo":1530,"voteCount":8126,"intelligenceIndex":97.5,"sweBenchScore":79.2,"gpqaScore":83.1,"mmluProScore":91,"livebenchScore":85.5,"scaleSealScore":97.4,"ifEvalScore":94,"outputSpeedTps":120,"latencyTtftMs":290,"qualitySpeedRatio":118,"contextWindowTokens":500000,"architectureType":"MoE","parameters":"Coding-Optimized","inputPricePerM":3,"outputPricePerM":15,"blendedPricePerM":6,"cacheReadPricePerM":0.3,"releaseDate":"2026-09","description":"Rank #3 Coding Champion worldwide (1774 Webdev Elo) and frontier agent tool orchestrator with blazing 120 tps output.","skillsTopic":"claude","docsUrl":"https://yakaai.com/doc/anthropics/anthropic-sdk-typescript","rankComposite":13},{"id":"gpt-5-6-sol-xhigh","name":"GPT-5.6 Sol (xHigh)","creator":"OpenAI","creatorSlug":"openai","licenseType":"proprietary","category":"reasoning","arenaElo":1457,"arenaCodingElo":1616,"arenaHardElo":1505,"arenaAgentElo":1520,"voteCount":41803,"intelligenceIndex":97.6,"sweBenchScore":76.8,"gpqaScore":83.1,"mmluProScore":91.8,"livebenchScore":85.1,"scaleSealScore":97.4,"ifEvalScore":93.9,"outputSpeedTps":96,"latencyTtftMs":380,"qualitySpeedRatio":94,"contextWindowTokens":400000,"architectureType":"Hybrid Reasoning","parameters":"Omni-MoE","inputPricePerM":4.5,"outputPricePerM":22.5,"blendedPricePerM":9,"cacheReadPricePerM":0.45,"releaseDate":"2026-07","description":"OpenAI flagship frontier reasoning engine with adaptive test-time compute, autonomous tool orchestration, and native multimodal synthesis.","skillsTopic":"openai","docsUrl":"https://yakaai.com/doc/openai/openai-node","rankComposite":14},{"id":"deepseek-v4-pro-high","name":"DeepSeek-V4 Pro (High)","creator":"DeepSeek","creatorSlug":"deepseek","licenseType":"open-source","category":"reasoning","arenaElo":1451,"arenaCodingElo":1582,"arenaHardElo":1472,"arenaAgentElo":1480,"voteCount":57564,"intelligenceIndex":93.8,"sweBenchScore":74.2,"gpqaScore":80.8,"mmluProScore":89.2,"livebenchScore":82.4,"scaleSealScore":94.5,"ifEvalScore":93.4,"outputSpeedTps":78,"latencyTtftMs":410,"qualitySpeedRatio":73,"contextWindowTokens":256000,"architectureType":"MoE","parameters":"750B (40B active)","minVramGb":64,"inputPricePerM":0.35,"outputPricePerM":1.4,"blendedPricePerM":0.61,"releaseDate":"2026-08","description":"State-of-the-art MIT open-weights reasoning model with Multi-head Latent Attention v2, DualPipe 2 inference, and native FP8 kernel optimization.","hfModelId":"deepseek-ai/DeepSeek-V4-Pro","ollamaModelId":"deepseek-v4:pro","skillsTopic":"deepseek","docsUrl":"https://yakaai.com/doc/deepseek-ai/DeepSeek-V3","rankComposite":15},{"id":"qwen3-8-27b","name":"Qwen 3.8 27B Instruct","creator":"Alibaba Qwen","creatorSlug":"qwen","licenseType":"open-source","category":"coding","arenaElo":1440,"arenaCodingElo":1593,"arenaHardElo":1445,"voteCount":25103,"intelligenceIndex":90.8,"sweBenchScore":67.5,"gpqaScore":75,"mmluProScore":84.8,"livebenchScore":75.1,"scaleSealScore":89.5,"ifEvalScore":92.4,"outputSpeedTps":135,"latencyTtftMs":210,"qualitySpeedRatio":123,"contextWindowTokens":128000,"architectureType":"Dense","parameters":"27.4B Dense","minVramGb":16,"inputPricePerM":0.1,"outputPricePerM":0.3,"blendedPricePerM":0.15,"releaseDate":"2026-08","description":"Single-GPU developer powerhouse fitting in 16GB VRAM. Exceeds GPT-4o on HumanEval and live coding evaluations.","hfModelId":"Qwen/Qwen3.8-27B","ollamaModelId":"qwen3.8:27b","skillsTopic":"qwen","docsUrl":"https://yakaai.com/doc/QwenLM/Qwen2.5","rankComposite":16},{"id":"qwen3-5-397b-a17b","name":"Qwen 3.5 397B-A17B","creator":"Alibaba Qwen","creatorSlug":"qwen","licenseType":"open-source","category":"llm","arenaElo":1437,"arenaCodingElo":1400,"arenaHardElo":1450,"voteCount":95123,"intelligenceIndex":91.5,"sweBenchScore":68.9,"gpqaScore":76.5,"mmluProScore":85.9,"livebenchScore":76.4,"scaleSealScore":90.8,"ifEvalScore":91,"outputSpeedTps":95,"latencyTtftMs":310,"qualitySpeedRatio":87,"contextWindowTokens":256000,"architectureType":"MoE","parameters":"397B (17B active)","minVramGb":32,"inputPricePerM":0.18,"outputPricePerM":0.54,"blendedPricePerM":0.27,"releaseDate":"2026-05","description":"Ultra-efficient Apache 2.0 MoE activating only 17B parameters per token, runnable on dual RTX 4090 GPUs at 95 tps with AWQ quantization.","hfModelId":"Qwen/Qwen3.5-397B-A17B","ollamaModelId":"qwen3.5:397b","skillsTopic":"qwen","docsUrl":"https://yakaai.com/doc/QwenLM/Qwen2.5","rankComposite":17},{"id":"gemma-4-26b-a4b","name":"Gemma 4 26B-A4B","creator":"Google DeepMind","creatorSlug":"google","licenseType":"open-source","category":"llm","arenaElo":1434,"arenaCodingElo":1359,"arenaHardElo":1438,"voteCount":6090,"intelligenceIndex":89.7,"sweBenchScore":65.8,"gpqaScore":73.9,"mmluProScore":83.7,"livebenchScore":73.8,"scaleSealScore":88.2,"ifEvalScore":89.5,"outputSpeedTps":155,"latencyTtftMs":180,"qualitySpeedRatio":139,"contextWindowTokens":128000,"architectureType":"MoE","parameters":"26B (4B active)","minVramGb":12,"inputPricePerM":0.08,"outputPricePerM":0.24,"blendedPricePerM":0.12,"releaseDate":"2026-07","description":"Ultra-compact edge MoE model activating only 4B parameters, designed for MacBook on-device inference, WebLLM, and local IDE co-pilots.","hfModelId":"google/gemma-4-26b-a4b","ollamaModelId":"gemma4:26b","skillsTopic":"gemini","docsUrl":"https://yakaai.com/doc/google-gemini/generative-ai-js","rankComposite":18},{"id":"deepseek-v4-flash","name":"DeepSeek-V4 Flash","creator":"DeepSeek","creatorSlug":"deepseek","licenseType":"open-source","category":"llm","arenaElo":1432,"arenaCodingElo":1582,"arenaHardElo":1435,"voteCount":51959,"intelligenceIndex":89.2,"sweBenchScore":64.9,"gpqaScore":73.1,"mmluProScore":83,"livebenchScore":72.9,"scaleSealScore":87.5,"ifEvalScore":89,"outputSpeedTps":180,"latencyTtftMs":160,"qualitySpeedRatio":161,"contextWindowTokens":256000,"architectureType":"MoE","parameters":"128B (12B active)","minVramGb":24,"inputPricePerM":0.08,"outputPricePerM":0.24,"blendedPricePerM":0.12,"releaseDate":"2026-07","description":"High-speed MIT open-weights workhorse delivering 180 tps inference with native 256k context and zero license restrictions.","hfModelId":"deepseek-ai/DeepSeek-V4-Flash","ollamaModelId":"deepseek-v4:flash","skillsTopic":"deepseek","docsUrl":"https://yakaai.com/doc/deepseek-ai/DeepSeek-V3","rankComposite":19},{"id":"mistral-large-3","name":"Mistral Large 3","creator":"Mistral AI","creatorSlug":"mistral","licenseType":"open-source","category":"llm","arenaElo":1429,"arenaCodingElo":1230,"arenaHardElo":1430,"voteCount":85822,"intelligenceIndex":88.5,"sweBenchScore":63.8,"gpqaScore":72.4,"mmluProScore":82.1,"livebenchScore":71.5,"scaleSealScore":86.8,"ifEvalScore":92.1,"outputSpeedTps":82,"latencyTtftMs":380,"qualitySpeedRatio":73,"contextWindowTokens":256000,"architectureType":"Dense","parameters":"140B Dense","minVramGb":48,"inputPricePerM":0.5,"outputPricePerM":1.5,"blendedPricePerM":0.75,"releaseDate":"2026-06","description":"European flagship fully open-sourced under Apache 2.0 with native French, German, Spanish, and multilingual reasoning.","hfModelId":"mistralai/Mistral-Large-3","ollamaModelId":"mistral-large:3","skillsTopic":"mistral","docsUrl":"https://yakaai.com/doc/mistralai/mistral-common","rankComposite":20},{"id":"deepseek-r1","name":"DeepSeek-R1 (RL)","creator":"DeepSeek","creatorSlug":"deepseek","licenseType":"open-source","category":"reasoning","arenaElo":1428,"arenaCodingElo":1465,"arenaHardElo":1460,"voteCount":18171,"intelligenceIndex":92.1,"sweBenchScore":70.2,"gpqaScore":78.4,"mmluProScore":86.8,"livebenchScore":75.9,"scaleSealScore":92.8,"ifEvalScore":90.2,"outputSpeedTps":62,"latencyTtftMs":650,"qualitySpeedRatio":57,"contextWindowTokens":128000,"architectureType":"Hybrid Reasoning","parameters":"671B (37B active)","minVramGb":48,"inputPricePerM":0.55,"outputPricePerM":2.19,"blendedPricePerM":0.96,"releaseDate":"2025-01","description":"Historical milestone in pure reinforcement learning reasoning, matching OpenAI o1 in formal logic, AIME mathematics, and algorithm synthesis.","hfModelId":"deepseek-ai/DeepSeek-R1","ollamaModelId":"deepseek-r1:latest","skillsTopic":"deepseek","docsUrl":"https://yakaai.com/doc/deepseek-ai/DeepSeek-V3","rankComposite":21},{"id":"codestral-2501","name":"Codestral 2501","creator":"Mistral AI","creatorSlug":"mistral","licenseType":"open-weights","category":"coding","arenaElo":1425,"arenaCodingElo":1450,"arenaHardElo":1415,"voteCount":38000,"intelligenceIndex":84.8,"sweBenchScore":59.4,"gpqaScore":67.2,"mmluProScore":79.5,"livebenchScore":68.2,"scaleSealScore":84,"ifEvalScore":89.2,"outputSpeedTps":125,"latencyTtftMs":260,"qualitySpeedRatio":106,"contextWindowTokens":256000,"architectureType":"Dense","parameters":"24B Dense","minVramGb":16,"inputPricePerM":0.25,"outputPricePerM":0.75,"blendedPricePerM":0.38,"releaseDate":"2025-01","description":"Specialized 256K Fill-In-the-Middle (FIM) code model optimized for low-latency auto-complete and multi-file code editing.","hfModelId":"mistralai/Codestral-22B-v0.1","ollamaModelId":"codestral:latest","skillsTopic":"mistral","docsUrl":"https://yakaai.com/doc/mistralai/mistral-common","rankComposite":22},{"id":"qwen-2-5-coder-32b-instruct","name":"Qwen 2.5 Coder 32B","creator":"Alibaba Qwen","creatorSlug":"qwen","licenseType":"open-source","category":"coding","arenaElo":1230,"arenaCodingElo":1465,"arenaHardElo":1420,"voteCount":5432,"intelligenceIndex":86.5,"sweBenchScore":61.2,"gpqaScore":69.8,"mmluProScore":82.1,"livebenchScore":70.8,"scaleSealScore":86.8,"ifEvalScore":90.4,"outputSpeedTps":120,"latencyTtftMs":290,"qualitySpeedRatio":104,"contextWindowTokens":128000,"architectureType":"Dense","parameters":"32.5B Dense","minVramGb":20,"inputPricePerM":0.12,"outputPricePerM":0.36,"blendedPricePerM":0.18,"releaseDate":"2024-11","description":"Proven Apache 2.0 code generator capable of full repository code understanding on a single 24GB RTX 3090/4090 GPU.","hfModelId":"Qwen/Qwen2.5-Coder-32B-Instruct","ollamaModelId":"qwen2.5-coder:32b","skillsTopic":"qwen","docsUrl":"https://yakaai.com/doc/QwenLM/Qwen2.5","rankComposite":23},{"id":"bge-en-icl","name":"BAAI BGE-en-ICL","creator":"BAAI","creatorSlug":"baai","licenseType":"open-source","category":"embedding","mtebRetrievalScore":76.5,"mtebRerankScore":73.8,"intelligenceIndex":94,"outputSpeedTps":1250,"latencyTtftMs":35,"qualitySpeedRatio":94,"contextWindowTokens":32768,"architectureType":"Bi-Encoder","parameters":"7B Dense","minVramGb":8,"inputPricePerM":0,"outputPricePerM":0,"blendedPricePerM":0,"releaseDate":"2025-02","description":"Number 1 overall ranking on MTEB Retrieval benchmark, supporting in-context learning with query-specific instruction prefixes.","hfModelId":"BAAI/bge-en-icl","rankComposite":24},{"id":"voyage-3-large","name":"Voyage 3 Large","creator":"Voyage AI","creatorSlug":"voyage","licenseType":"proprietary","category":"embedding","mtebRetrievalScore":75.8,"mtebRerankScore":72.9,"intelligenceIndex":93.2,"outputSpeedTps":2400,"latencyTtftMs":28,"qualitySpeedRatio":93,"contextWindowTokens":32000,"architectureType":"Bi-Encoder","parameters":"1024-dim","inputPricePerM":0.12,"outputPricePerM":0,"blendedPricePerM":0.12,"releaseDate":"2025-01","description":"Commercial embedding champion engineered by Stanford researchers, optimized for technical documents, legal contracts, and financial RAG.","rankComposite":25},{"id":"bge-reranker-v2-m3","name":"BAAI BGE-Reranker-v2-m3","creator":"BAAI","creatorSlug":"baai","licenseType":"open-source","category":"reranker","mtebRerankScore":75.2,"intelligenceIndex":92.5,"outputSpeedTps":850,"latencyTtftMs":45,"qualitySpeedRatio":92,"contextWindowTokens":8192,"architectureType":"Cross-Encoder","parameters":"567M Cross","minVramGb":4,"inputPricePerM":0,"outputPricePerM":0,"blendedPricePerM":0,"releaseDate":"2024-11","description":"Open-source cross-encoder reranker for high-precision top-K candidate re-scoring in production 2-stage RAG pipelines.","hfModelId":"BAAI/bge-reranker-v2-m3","rankComposite":26},{"id":"cohere-rerank-v3-5","name":"Cohere Rerank 3.5","creator":"Cohere","creatorSlug":"cohere","licenseType":"proprietary","category":"reranker","mtebRerankScore":75,"intelligenceIndex":92,"outputSpeedTps":1100,"latencyTtftMs":40,"qualitySpeedRatio":92,"contextWindowTokens":4096,"architectureType":"Cross-Encoder","parameters":"Cross-Encoder","inputPricePerM":1,"outputPricePerM":0,"blendedPricePerM":1,"releaseDate":"2024-11","description":"State-of-the-art multilingual reranking API with native support for structured JSON fields, code snippets, and semi-structured tables.","rankComposite":27},{"id":"text-embedding-3-large","name":"text-embedding-3-large","creator":"OpenAI","creatorSlug":"openai","licenseType":"proprietary","category":"embedding","mtebRetrievalScore":71.2,"mtebRerankScore":68.5,"intelligenceIndex":88,"outputSpeedTps":3200,"latencyTtftMs":22,"qualitySpeedRatio":88,"contextWindowTokens":8191,"architectureType":"Bi-Encoder","parameters":"3072-dim","inputPricePerM":0.13,"outputPricePerM":0,"blendedPricePerM":0.13,"releaseDate":"2024-01","description":"Most widely deployed enterprise embedding model supporting Matryoshka dimension truncation (down to 256 or 1024 dimensions) to reduce vector storage costs.","skillsTopic":"openai","docsUrl":"https://yakaai.com/doc/openai/openai-node","rankComposite":28},{"id":"snowflake-arctic-embed-m-v2","name":"Snowflake Arctic Embed M v2","creator":"Snowflake","creatorSlug":"snowflake","licenseType":"open-source","category":"embedding","mtebRetrievalScore":72.4,"mtebRerankScore":69.1,"intelligenceIndex":89.4,"outputSpeedTps":4800,"latencyTtftMs":15,"qualitySpeedRatio":89,"contextWindowTokens":8192,"architectureType":"Bi-Encoder","parameters":"305M Dense","minVramGb":2,"inputPricePerM":0,"outputPricePerM":0,"blendedPricePerM":0,"releaseDate":"2024-12","description":"Apache 2.0 embedding model matching Cohere and OpenAI quality while being 3x faster and ultra-lightweight for serverless execution.","hfModelId":"Snowflake/snowflake-arctic-embed-m-v2.0","rankComposite":29}],"categories":[{"id":"all","name":"All Models","count":29},{"id":"reasoning","name":"🧠 Frontier Reasoning","count":11},{"id":"agent","name":"🤖 Autonomous Agents","count":11},{"id":"coding","name":"💻 Coding Champions","count":20},{"id":"efficiency","name":"⚡ Speed & Ratio","count":29},{"id":"llm","name":"⚡ High-Throughput LLMs","count":7},{"id":"embedding","name":"🔍 Embeddings & Rerankers","count":6}],"creators":[{"id":"all","name":"All Creators","count":29},{"id":"google","name":"Google DeepMind","count":4},{"id":"anthropic","name":"Anthropic","count":4},{"id":"xiaomi","name":"Xiaomi","count":1},{"id":"qwen","name":"Alibaba Qwen","count":4},{"id":"meta","name":"Meta","count":1},{"id":"xai","name":"xAI","count":1},{"id":"moonshot","name":"Moonshot AI","count":1},{"id":"zhipu","name":"Zhipu AI","count":1},{"id":"openai","name":"OpenAI","count":2},{"id":"deepseek","name":"DeepSeek","count":3},{"id":"mistral","name":"Mistral AI","count":2},{"id":"baai","name":"BAAI","count":2},{"id":"voyage","name":"Voyage AI","count":1},{"id":"cohere","name":"Cohere","count":1},{"id":"snowflake","name":"Snowflake","count":1}]}