{"generatedAt":"2026-09-26T06:03:03.526Z","scope":"Decisions","level":"category","scopeBoards":1,"scopeReports":["LargitData — Jev on multi-turn RAG routing"],"scopeNote":"Accuracy and calibration only. JevBench also ranks speed and price, which this index does not read as capability, so a large API model can lead this column at a cost a router would not pay.","minBoards":2,"reports":false,"total":5,"scopeTotal":85,"models":[{"rank":13,"id":"deepseek-v4-1-flash","name":"DeepSeek V4.1 Flash","org":"DeepSeek","family":"DeepSeek","index":62.4,"estimatedIndex":null,"scopeScore":73.3,"ranked":true,"measuredWeight":0.7,"rankedInScope":true,"position":1,"size":{"paramsB":763.2,"source":"huggingface","label":"763B"},"sizeTier":"large","covered":8,"coverable":19,"reports":1,"agreement":"mixed","access":{"recommended":{"kind":"openrouter","name":"OpenRouter","url":"https://openrouter.ai/deepseek/deepseek-v4.1-flash","model":"deepseek/deepseek-v4.1-flash","baseUrl":"https://openrouter.ai/api/v1","note":"OpenAI-compatible chat completions; one key for every listed model."},"alternatives":[{"kind":"vendor","name":"DeepSeek API","url":"https://api-docs.deepseek.com","note":"The vendor’s own developer site; first-party features and pricing."},{"kind":"weights","name":"Hugging Face","url":"https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash","model":"deepseek-ai/DeepSeek-V4.1-Flash","note":"Open weights; run with vLLM, SGLang or similar."}],"rule":"Recommended channel: the OpenRouter id through its OpenAI-compatible endpoint — one key, every listed model, ids stable across vendors. Use the vendor’s own API for first-party features, and the weights for self-hosting."}},{"rank":30,"id":"gpt-6-luna","name":"GPT-6 Luna","org":"OpenAI","family":"GPT","index":55.7,"estimatedIndex":null,"scopeScore":72.7,"ranked":true,"measuredWeight":0.85,"rankedInScope":true,"position":2,"size":null,"sizeTier":"undisclosed","covered":9,"coverable":19,"reports":0,"agreement":"mixed","access":{"recommended":{"kind":"openrouter","name":"OpenRouter","url":"https://openrouter.ai/openai/gpt-6-luna","model":"openai/gpt-6-luna","baseUrl":"https://openrouter.ai/api/v1","note":"OpenAI-compatible chat completions; one key for every listed model."},"alternatives":[{"kind":"vendor","name":"OpenAI Platform","url":"https://platform.openai.com/docs","note":"The vendor’s own developer site; first-party features and pricing."}],"rule":"Recommended channel: the OpenRouter id through its OpenAI-compatible endpoint — one key, every listed model, ids stable across vendors. Use the vendor’s own API for first-party features, and the weights for self-hosting."}},{"rank":41,"id":"gpt-5-6-luna","name":"GPT-5.6 Luna","org":"OpenAI","family":"GPT","index":52.9,"estimatedIndex":null,"scopeScore":70.6,"ranked":true,"measuredWeight":0.75,"rankedInScope":true,"position":3,"size":null,"sizeTier":"undisclosed","covered":11,"coverable":19,"reports":1,"agreement":"mixed","access":{"recommended":{"kind":"openrouter","name":"OpenRouter","url":"https://openrouter.ai/openai/gpt-5.6-luna","model":"openai/gpt-5.6-luna","baseUrl":"https://openrouter.ai/api/v1","note":"OpenAI-compatible chat completions; one key for every listed model."},"alternatives":[{"kind":"vendor","name":"OpenAI Platform","url":"https://platform.openai.com/docs","note":"The vendor’s own developer site; first-party features and pricing."}],"rule":"Recommended channel: the OpenRouter id through its OpenAI-compatible endpoint — one key, every listed model, ids stable across vendors. Use the vendor’s own API for first-party features, and the weights for self-hosting."}},{"rank":null,"id":"gemma-4-31b","name":"Gemma 4 31B","org":"Google","family":"Gemma","index":null,"estimatedIndex":{"score":52.4,"measures":6,"anchors":57,"bound":null},"scopeScore":60.2,"ranked":false,"measuredWeight":0.25,"rankedInScope":false,"position":null,"size":{"paramsB":31.3,"source":"huggingface","label":"31B"},"sizeTier":"medium","covered":7,"coverable":19,"reports":2,"agreement":"mixed","access":{"recommended":{"kind":"openrouter","name":"OpenRouter","url":"https://openrouter.ai/google/gemma-4-31b-it","model":"google/gemma-4-31b-it","baseUrl":"https://openrouter.ai/api/v1","note":"OpenAI-compatible chat completions; one key for every listed model."},"alternatives":[{"kind":"vendor","name":"Gemini API","url":"https://ai.google.dev/gemini-api/docs","note":"The vendor’s own developer site; first-party features and pricing."},{"kind":"weights","name":"Hugging Face","url":"https://huggingface.co/google/gemma-4-31B-it","model":"google/gemma-4-31B-it","note":"Open weights; run with vLLM, SGLang or similar."}],"rule":"Recommended channel: the OpenRouter id through its OpenAI-compatible endpoint — one key, every listed model, ids stable across vendors. Use the vendor’s own API for first-party features, and the weights for self-hosting."}},{"rank":null,"id":"gemini-3-1-flash-lite","name":"Gemini 3.1 Flash Lite","org":"Google","family":"Gemini","index":null,"estimatedIndex":{"score":51.7,"measures":20,"anchors":57,"bound":null},"scopeScore":54.1,"ranked":false,"measuredWeight":0.25,"rankedInScope":true,"position":25,"size":null,"sizeTier":"undisclosed","covered":7,"coverable":19,"reports":0,"agreement":"mixed","access":{"recommended":{"kind":"openrouter","name":"OpenRouter","url":"https://openrouter.ai/google/gemini-3.1-flash-lite","model":"google/gemini-3.1-flash-lite","baseUrl":"https://openrouter.ai/api/v1","note":"OpenAI-compatible chat completions; one key for every listed model."},"alternatives":[{"kind":"vendor","name":"Gemini API","url":"https://ai.google.dev/gemini-api/docs","note":"The vendor’s own developer site; first-party features and pricing."}],"rule":"Recommended channel: the OpenRouter id through its OpenAI-compatible endpoint — one key, every listed model, ids stable across vendors. Use the vendor’s own API for first-party features, and the weights for self-hosting."}}],"note":"Scores are 0–100 standardized across the models each source lists; 50 is that measure’s average, not a grade. Overall ranks a model once boards from two independent publishers have scored it; a category or domain ranks any model a recognised board measured there. Report (✱) figures never rank a model on their own. This is a snapshot dated generatedAt, not a live feed; every figure links to its publisher."}