MLExpert VRAM Requirements

MLExpert at a glance

Made by

Shanghai AI Lab, China

Released

October 1, 2026

Parameters

34.7B, 2.9B active

Context window

256K tokens

License

Apache 2.0, commercial use allowed

Reads

Text, images and video

Tool calling

Supported

Reasoning

Thinking mode you can switch on or off

Paper

MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering (opens in a new tab) (2026)

On Hugging Face

16 likes, 630 downloads last month (opens in a new tab)

Runs with

vLLM, SGLang, llama.cpp (Ollama, LM Studio), MLX (Mac)

VRAM by precision and context length

Weights plus the KV cache of one request (BF16 cache) and the model's fixed recurrent state, in GiB. Add 1 to 3 GiB for the inference engine itself.

MLExpert GGUF files

Exact size of each quantization in bartowski/InternScience_MLExpert-GGUF (opens in a new tab), and the smallest GPU that runs it with llama.cpp, Ollama or LM Studio and an 8K-token context.

Read every day from Hugging Face. Split files are added together; vision projectors are left out.

How many GPUs to run MLExpert

Fewest GPUs that fit the weights plus one 8K-token request. Data-center GPUs use vLLM (90% of memory usable), the others llama.cpp. Several GPUs split the model with tensor or pipeline parallelism.

8-bit: FP8 with vLLM (INT8 on GPUs without FP8), Q8_0 with llama.cpp. 4-bit: AWQ / GPTQ with vLLM, Q4_K_M with llama.cpp, both counted at the Q4_K_M size.

Cheapest way to run MLExpert today

For each GPU: the fewest cards that fit MLExpert with one 8K-token request, the most faithful weight format at that count, and the cheapest on-demand price (2026-10-23).

On-demand prices from AWS, Google Cloud, Azure, RunPod, Vast.ai, Oracle Cloud, OVHcloud, Scaleway, Verda, Akash and UpCloud, checked every hour. Serving many users needs more KV cache, so more memory: size it in the calculator. Some links are affiliate links: StudioTV may earn a commission, at no extra cost to you.

AWS, Google Cloud and UpCloud prices: Spare Cores (opens in a new tab), CC BY-SA 4.0 (opens in a new tab), shared under the same license.

Recommended settings for MLExpert

Sampling defaults from the official generation_config.json (opens in a new tab) of InternScience/MLExpert.

vLLM and SGLang apply these defaults unless a request sets its own. With llama.cpp, pass them yourself: --temp 1 --top-p 0.95 --top-k 20 --min-p 0

Reasoning. Thinking mode you can switch on or off. With vLLM or SGLang, set it per request: "chat_template_kwargs": {"enable_thinking": false} turns it off, true on. Reasoning tokens take context and KV cache like any other token: leave room for them.

KV cache: how memory grows with context

Architecture: 40 layers · 10 full attention (2 KV × 256) · 30 linear attention. Its KV cache is 4.0× smaller than if every layer used full attention, because only 10 of its 40 layers keep a KV cache; the others hold a small fixed-size state. Only 2.9B parameters are active per token, but all 35B must sit in memory: VRAM depends on the total, speed on the active part.

Plus 61.4 MiB of recurrent state per request, whatever the context.

MLExpert VRAM badge

For a model card or a README: the 4-bit size and the smallest GPU it fits on, linked to this page. Data as JSON: /api/models/mlexpert.json.

[![MLExpert VRAM](https://studiotvai.com/badge/mlexpert.svg)](https://studiotvai.com/vram-requirements/mlexpert)

Other models

Same family (hybrid linear attention): Qwen3.5 4B · MiniCPM-V-4.7 1B · Raon-1 mini preview · Qwen3.5 9B · Qwen3.8 27B · Qwen3.6 35B-A3B · Qwen3.5 122B-A10B · Qwen3.8-Flash-Next · LFM2.5 8B-A1B · LFM2 24B-A2B · Ornith 1.5 35B-A3B · Ornith 1.5 9B · Clef · Clef Flash · D1 3B · JEV 27B VL · LightOnOCR-3 4B · Agnes-3.0-Qwen · Agnes-3.0-Flash · Qwen3.6 27B · Qwen3.5 27B · Qwen3.5 35B-A3B · MiniCPM-V-4.7 35B-A3B · LFM2.5 24B-A2B · Ornith-1.5 122B · Sage-A1.5 · Agents-A1 4B · Ling-3.0-flash-Sante

Estimates, not guarantees: computed from the official config.json with the same engine as the LLM VRAM Calculator. Real usage depends on your engine version and settings.