NVIDIA Nemotron 3.5 Super VL 對決 MiniCPM-V 4.7 35B-A3B：口袋裡的 3 GB 對上機架上的 4× H100

One of these runs on an iPhone. The other runs on four H100s. That is the whole comparison, and everything else is detail. MiniCPM-V 4.7 35B-A3B is OpenBMB's first Mixture-of-Experts entry in the MiniCPM-V line, and the model card's claim is specific enough to check: more than 13 tokens per second on an iPhone 17 Pro Max using roughly 3 GB of memory, with the run described as smooth rather than as a demo. N​VIDIA N​emotron 3.5 Super VL is a 120B-total, 12B-active vision-language model whose card puts the deployment envelope at 4× H100 or 2× GB200. Both are open weights. Both are vision-language models. They are not options for the same decision.

無論如何，它們值得拿來比較的原因在於：它們在機制上看法一致，在規模上卻各執一詞。兩者都是混合專家（Mixture-of-Experts）模型，設計宗旨是以總體規模換取每 token 成本。兩者都會在語言模型必須費心處理之前，先壓縮視覺 token。兩者都押注視覺語言領域的未來，會是許多模型分散在許多地方，而非一個大型模型躲在單一 API 之後。只是它們下的賭注，在硬體上相差三個數量級。

主動參數巧合

N​VIDIA N​emotron 3.5 Super VL is 120B total with 12B active per token — a Mamba-2–Transformer hybrid MoE with a C-RADIO vision encoder and 512 routed experts per MoE layer, holding up to 1M tokens of context. MiniCPM-V 4.7 35B-A3B is 35B total with 3B active. Different by a factor of four in total size and four in active compute, but the same architectural instinct: keep the parameter count high for knowledge and the per-token cost low for speed.

OpenBMB 對這個問題的貢獻是 Lookahead Expert Scheduling（前瞻專家排程），正是它讓一個 3B 活躍參數的模型能讓專家常駐，而不致讓記憶體匯流排陷入顛簸——這也是它能在 3 GB 中運行，而不需要一台工作站的原因。其實作位於一個分支 llama.cpp-omni，而附帶於它的說法相當謹慎：第一個能在裝置上順暢運行、且不犧牲效能的開源 MoE 模型。「不犧牲」能否經得起與 120B 模型相比，這張卡並未觸及，也不該觸及——一支能在手機上運行的模型，在準確度上並不是在跟一支需要四張 H100 的模型競爭。它競爭的，是自己究竟能不能存在。

• Total / active parameters — N​emotron 3.5 Super VL: 120B / 12B. MiniCPM-V 4.7 35B-A3B: 35B / 3B.

• Deployment target — N​emotron 3.5 Super VL: 4× H100 or 2× GB200, served via vLLM or SGLang. MiniCPM-V 4.7 35B-A3B: phone, laptop, edge box — iOS, Android and HarmonyOS are all named.

• Memory at the smallest end — N​emotron 3.5 Super VL: not applicable. MiniCPM-V 4.7 35B-A3B: about 3 GB.

• Context — N​emotron 3.5 Super VL: up to 1M tokens, video up to two minutes. MiniCPM-V 4.7 35B-A3B: not the model's selling point; the card leads on footprint and speed.

• Licence — N​emotron 3.5 Super VL: OpenMDW-1.1. MiniCPM-V 4.7 35B-A3B: Apache-2.0.

對「token 很貴」的兩種不同答案

兩支團隊都著手處理同一個瓶頸——一張圖片所耗費的視覺 token 數量——最終卻抵達了不同的地方。

MiniCPM-V 4.7 35B-A3B 是在視覺編碼器內部完成這件事的，其早期壓縮技術源自 LLaVA-UHD v4，接著再視影像需求，混用 4× 與 16× 的視覺 token 壓縮。精細的細節走昂貴的路徑；一張收據則走便宜的路徑。壓縮是在語言模型涉入之前就發生，這正是讓裝置端數值得以實現的原因。

N​emotron 3.5 Super VL does not lead with token compression at all. Its answer is context: 1M tokens, so you can afford to keep the detail rather than throw it away, and a hybrid Mamba-2 layer to keep the cost of a very long sequence from exploding. The model's own video path is configured generously by default — 2 FPS and 256 frames, dropping to 1 FPS and 128 frames at 1080p — which is a sampling answer, not a compression answer.

那些是不相容的哲學，而兩者在各自的部署情境下都是正確的。如果你有 3 GB，你就壓縮。如果你有 4× H100，你就保留細節，並為上下文付出代價。

數字能告訴你什麼，不能告訴你什麼

以下是對雙方立場的誠實說明。

N​emotron 3.5 Super VL's benchmark table is broad and vendor-collected through NeMo Gym: GPQA-Diamond 89.52, SWE-bench Verified 71.10, MMLongBench-Doc 62.82, OCRBench v2 at 68.12 English and 59.15 Chinese, CharXiv reasoning 80.82, V* 85.60. N​VIDIA chose the baselines. None of it has been independently reproduced.

MiniCPM-V 4.7 35B-A3B 的模型卡是以廠商圖表而非數值表格來呈現其基準測試結果，因此沒有數字可供引用，也沒有基準清單可供檢視。該模型卡中最強烈的具體宣稱是比較性且方向性的——在 Instruct 模式下，以更少的推論時間取得比 Qwen3.6 35B-A3B 更高的整體分數——這是關於其自身量級內效率的宣稱，而非關於對抗一個規模達其十二倍的模型之能力的宣稱。

因此，這兩者在基準測試數據上的比較表並不存在，也無法建立。任何人提出這樣的表，都只是在把廠商圖表與廠商表格拼湊起來，然後把結果稱為測量。

比準確度更能決定專案成敗的部署細節

MiniCPM-V 4.7 35B-A3B 推出時即首日支援 Rockchip、Houmo、Intel、AMD 與 Arm 晶片，並支援 SGLang、vLLM、llama.cpp 與 Ollama，以及 GGUF、BNB、AWQ、GPTQ 與 FP8 量化。它也整合 SWIFT 與 LLaMA-Factory 以進行微調。那份清單才是真正的產品。能在你已經購入的加速器上執行、並採用你執行環境已支援的量化格式的模型，才是你能在本季出貨的模型。

N​emotron 3.5 Super VL's runtime story is narrower and deeper: vLLM at vllm/vllm-openai:v0.32.0 and SGLang at lmsysorg/sglang:v0.5.20, with EAGLE speculative decoding at topk 1 and Multi-Token Prediction built into the architecture. Two runtimes, both first-class, both specified to the exact version. That is what a datacenter deployment wants — a validated path, not a menu.

If you are choosing hardware to fit the model, take MiniCPM-V 4.7 35B-A3B. If you are choosing the model to fit hardware you already have, take N​emotron 3.5 Super VL and follow its card's versions exactly.

同時運行兩者的理由

The pattern that shows up in real products is not a choice between these two. It is a small VLM on the device handling the work that has to be instant and local — reading a form, describing a scene, filtering out the frames nobody cares about — and a large VLM in the datacenter handling the work that has to be right. MiniCPM-V 4.7 35B-A3B is explicitly designed for the first role. N​emotron 3.5 Super VL is designed for the second.

That architecture has a cost, and the cost is that you are now running two model stacks and a text stack, with three sets of credentials and three sets of failure modes. OrcaRouter is the part that removes the second problem: 207 models behind one key at provider list price with zero markup, so a vendor price change lands on your bill the same day, and automatic failover when an upstream degrades. Neither N​VIDIA N​emotron 3.5 Super VL nor MiniCPM-V 4.7 35B-A3B is in that catalogue — both are open weights you host yourself, which is the point of choosing either. What one key gets you is the surrounding text and code models, so the pipeline does not need a new contract every time you add a stage.

哪一個，取決於像素所在的位置

若推論必須在資料所在之處進行——手機上、車輛內、工廠現場，或影像無法離開建築物的診所——或者你的使用者處於離線狀態，又或者託管 VLM 的每次呼叫成本正是扼殺單位經濟效益的元兇，那就選擇 MiniCPM-V 4.7 35B-A3B。其授權為 Apache-2.0，量化選項豐富，晶片支援清單也刻意拉得很長。

Take N​VIDIA N​emotron 3.5 Super VL if the work is long-context and reasoning-heavy: multi-page documents, mixed layouts, charts that need interpretation, two-minute video clips, agentic tool use. The 1M-token context and 12B active parameters are the reason it exists, and no on-device model substitutes for them.

如果你不確定，決定性的問題不是準確度，而是這張影像是否獲准離開裝置。這個答案會在參考任何基準測試之前，就先決定要選哪個模型。

MiniCPM-V 4.7 35B-A3B 的模型卡是裝置端相關宣稱的來源——每秒 13 個以上 token、約 3 GB 的佔用空間、首日晶片清單以及量化格式——而這些全都是該模型自行報告的結果。在你投入之前，請先閱讀晶片清單，因為其中的「支援」代表存在執行階段路徑，並不代表你特定的開發板已經過基準測試。

N​VIDIA's N​emotron developer page is where the 4× H100 / 2× GB200 guidance and the runtime versions are published, along with the link to the technical report behind the benchmark table. The version numbers in the card are exact for a reason: vLLM v0.32.0 and SGLang v0.5.20 with EAGLE topk 1 are the validated combination, and drifting from them is how a working deployment becomes a debugging session.

The larger point is that the vision-language field has stopped converging on one shape. OpenBMB is pushing a 3B-active model into pockets and N​VIDIA is pushing a 12B-active model across a million tokens of context, in the same month, under licences that both permit commercial use. Picking between them is picking a deployment, and the deployment is decided by where the pixels live.

基準測試會過時，價格會波動，但失敗紀錄不會：這一類模型背後有據可查的 AI 事故與代理失誤，全都收錄於the OrcaRouter AI Incident Archive，其中每一筆紀錄都附有自己的來源、嚴重程度、可信度評級與爭議標記。