selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

tencent logo

WeMM-Video-35BA3B

tencent·Instruct·released 17 Nov 2026

Apache 2.036.2BA5BMoE256K ctx ≈23 GB
Your GPU
hf download tencent/WeMM-Video-35BA3B
Setup ↓

WeMM-Video-35B-A3B is the mixture-of-experts (35B total, 3B active) checkpoint in the WeMM-Video family. WeMM-Video is a Chinese–English bilingual family for joint audio-visual understanding and reasoning. It accepts text, images, audio, and video and generates text responses.

Notify me when

How much VRAM does WeMM-Video-35BA3B need?

At Q4_K_M WeMM-Video-35BA3B needs about 23 GB with an 8K context (23 GB at Q4, 72 GB at full precision). Pick your GPU to see what fits and how fast it runs.

--vram
any
--gpus
--quant

Needs: ≈23 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.

Change quant, context or KV cache to compare
Advanced: context 8K · KV cache 16-bit
Context
KV cache

Smallest GPU for each quant and context (KV cache 16-bit)

Quantisation8K16K32K64K128K256K
Q3_K_M
Q4_K_M
Q5_K_M
Q6_K
Q8_0
FP16

≤16 GB card24–32 GB48–96 GBmore

QuantisationQualityWeightsKV cacheTotal
Q2_KHeavy lossSmallest, noticeably worse qualityHeavy loss11 GB 0.6 GB≈13 GB
Q3_K_MNoticeable lossSmall, some quality lossNoticeable loss16 GB 0.6 GB≈18 GB
Q4_K_MrecommendedSmall lossRecommended default for local useSmall loss20 GB 0.6 GB≈23 GB
Q5_K_MNear-losslessNear-lossless for most tasksNear-lossless24 GB 0.6 GB≈26 GB
Q6_KNear-losslessPractically losslessNear-lossless28 GB 0.6 GB≈30 GB
Q8_0LosslessLossless in practice, twice the size of Q4Lossless36 GB 0.6 GB≈39 GB
FP16 / BF16LosslessFull precision, as releasedLossless67 GB 0.6 GB≈72 GB

Estimates for 8K tokens of context: weights + KV cache + runtime overhead. All experts stay in memory, but only the active ones are read per token. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it.

How to pick a quant

Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.

  • LosslessF16, Q8_0
  • Near-losslessQ6_K, Q5_K_M
  • Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
  • Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
  • Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K

Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B

How much context fits?

Model quant
KV cache

At 8K context, Q4_K_M needs ≈23 GB(20 GB weights + 0.6 GB KV cache + 1.5 GB overhead). Pick a GPU above to see how much context fits.

010203040501K2K4K8K16K32K64K128K256KGB

Weights + overheadKV cache

Can the Tesla P40 run WeMM-Video-35BA3B? →

Speed vs quality per quant

Smaller quants run faster and fit in less memory, but lose quality.

Q2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0F16

Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.

Real-world speed of WeMM-Video-35BA3B

Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →

No medians yet. Be one of the first 3 to report a setup.

All reports

Run WeMM-Video-35BA3B locally

Quantised versions and downloads

About WeMM-Video-35BA3B

WeMM-Video-35B-A3B is the mixture-of-experts (35B total, 3B active) checkpoint in the WeMM-Video family. WeMM-Video is a Chinese–English bilingual family for joint audio-visual understanding and reasoning. It accepts text, images, audio, and video and generates text responses. • Weights · BF16 Safetensors · 67.39 GiB of weight files • Quantized variants · FP8 Block128 · AWQ W4A16 · GGUF Q4_K_M · NVFP4 W4A16 WeMM-Video can combine video frames with the original soundtrack to describe events, transcribe speech, and reason about what is seen and heard. It also accepts silent video, standalone audio, images, and text, with responses in Chinese or English. Vision and audio encoders turn media into tokens that the language model processes alongside text. Video frames carry timestamps. The family was pretrained on approximately 1.2 trillion multimodal tokens, then post-trained with multi-teacher OPD, anti-repetition MrDPO, and specialist reinforcement learning. See the project overview for family details. Set up the Transformers environment. From the WeMM-Video repository root, run the example on an MP4 with a soundtrack. The script loads the model and processor from the Hub and processes both the frames and the original audio track. Use --revision to pin the model and processor to a Hub commit. Size GPU memory for the full weight files, media inputs, and generation cache.

Full model card on Hugging Face →

Alternatives to WeMM-Video-35BA3B

All alternatives →
Qwen logo

Qwen3.6-35B-A3B

Qwen·Vision

≈22 GB

Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding…

36BA3B256K ctx Apache 2.0
3.8MApr 2026ollama
openbmb logo

MiniCPM-V-4.7-35B-A3B

openbmb·Vision

≈22 GB

GitHub · On-device results · Evaluation · Examples · Usage · Model weights

35.2BA3B256K ctx Apache 2.0
876.3KOct 2026gguf
Qwen logo

Qwen3.8-27B

Qwen·Vision

≈19 GB

Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date.

27.8B256K ctx Apache 2.0
6.5MAug 2026ollama
google logo

gemma-4-31B-it

google·Vision

≈27 GB

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

31.3B256K ctx Apache 2.0
6.1MMar 2026ollama
Qwen logo

Qwen3.5-35B-A3B

Qwen·Vision

≈22 GB

Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.

36BA3B256K ctx Apache 2.0
1.3MFeb 2026ollama
Qwen logo

Qwen3.6-27B

Qwen·Vision

≈19 GB

Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding…

27.8B256K ctx Apache 2.0
2.4MApr 2026ollama

WeMM-Video-35BA3B FAQ

How much VRAM does WeMM-Video-35BA3B need?
About 23 GB at Q4_K_M with an 8K context. Q8_0 needs about 39 GB and full precision about 72 GB. Longer contexts need more memory for the KV cache.
Can WeMM-Video-35BA3B run on the Tesla P40?
Yes. It fits at Q4_K_M (about 23 GB), at an estimated 63 tokens/s. It fits comfortably on the Radeon RX 7900 XTX, Instinct MI50 32GB, RTX 5090.
Can I use WeMM-Video-35BA3B commercially?
Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.
What is the context length of WeMM-Video-35BA3B?
262,144 tokens (256K). The full window needs far more memory than an 8K context.

Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.