{"id":"orvena-4b","slug":"orvena-4b","name":"Orvena 4B","orgId":"orvena","family":"orvena-4b","releaseDate":null,"availability":"ga","openWeights":true,"license":"Apache-2.0","paramsTotal":4000000000,"paramsSource":"slug","contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://huggingface.co/Orvena/Orvena-4B","hfRepo":null,"modalities":["text","image"],"aliases":["Orvena 4B","orvena-4b","orvena 4b","Orvena/Orvena-4B","Orvena/Orvena-4B-MLX-4bit","Orvena-4B-MLX-4bit"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-11-05","announced":"2026-11-04","blurb":"Orvena's open-weight on-device assistant model for the Orvena iPhone app: a LoRA fine-tune of Qwen3.5-4B (LoRA r=16, alpha 320 on all 248 linear layers, merged) that keeps the base architecture, including the vision encoder, and answers without a separate thinking step by default. Trained for phone tools (calendar, alarms, places, travel times, web) and MCP connections such as Notion, Todoist, Airtable and Gmail, in 30 languages. Apache-2.0 bf16 weights with a retrained MTP head, plus a 4-bit MLX build the app downloads. No hosted API stated.","api":"none","openWeights":"yes","params":{"text":"4B (Qwen3.5-4B backbone)","source":"Orvena's model card and X thread"},"facts":[{"label":"Base model","value":"Qwen/Qwen3.5-4B (Alibaba); 24 Gated DeltaNet layers, 8 full-attention layers, vision encoder retained"},{"label":"Training","value":"LoRA r=16, alpha 320 on all 248 language-model linear layers for one epoch, then merged; thinking off; MTP head retrained on the fine-tuned model's outputs"},{"label":"Training data","value":"3,397 simulated-iPhone scenarios (11,161 rounds) with Orvena's tool catalog and MCP tool lists from 12 services, plus no-tool/restraint sets and 600 IFEval-style instruction-following rows"},{"label":"App","value":"Default model in Orvena 2.5 for iPhone; 4-bit MLX build is what the app downloads"},{"label":"Languages","value":"30 languages listed on the model card"},{"label":"Checkpoints","value":"Orvena/Orvena-4B (bf16 + MTP) and Orvena/Orvena-4B-MLX-4bit on Hugging Face; technical report PDF on the card"},{"label":"License","value":"Apache-2.0"},{"label":"API","value":"No hosted API stated; vLLM / Transformers / SGLang instructions for local serving"}],"claims":["Lab claim: on the Orvena bench of 547 held-out scenarios, Orvena 4B passed 440 (80.4%), against 345 (63.1%) for Qwen3.5-4B and 104 (19.0%) for Apple's on-device model.","Lab claim: on multi-step tasks, 43 of 70 against 17 for the base model.","Lab claim: on an iPhone 17 Pro Max with the 4-bit MLX build, median time per task was 15.1 s (base thinking off: 16.3 s; thinking auto: 37.0 s)."],"sources":[{"kind":"repo","title":"Orvena/Orvena-4B model card (Hugging Face)","url":"https://huggingface.co/Orvena/Orvena-4B"},{"kind":"repo","title":"Orvena/Orvena-4B-MLX-4bit model card (Hugging Face)","url":"https://huggingface.co/Orvena/Orvena-4B-MLX-4bit"},{"kind":"lab","title":"Orvena announcement on X","url":"https://x.com/orvenaAI/status/2118321172420960536","date":"2026-11-05"},{"kind":"lab","title":"Orvena app site","url":"https://orvena.app"}],"firstBenchmark":null},"organization":{"id":"orvena","name":"Orvena","slug":"orvena","country":null,"url":"https://orvena.app","color":"#11110F","mark":"O"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Nov 4, 2026; no independent results yet.","lead":"As of Nov 22, 2026, Orvena 4B is pending benchmarks on isitsota: announced Nov 4, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}