WeMM Video 35BA3B

Language Tencent Partly verified
  • In Input modalities: TextImage
  • Tool use
  • Open weights

Prices

0 offerings · prices as listed

No priced offering is listed for this model yet.

Details

Licence
apache-2.0
Family
wemm-video

Run locally

Weights: tencent/WeMM-Video-35BA3B
  • Parameters 36.2B
  • Active 5B (estimated) · 256 experts, 8 per token
  • Native precision BF16
  • Layers 40
  • KV cache per token 20 KB
  • Config positions 256K (config.json max_position_embeddings)
My hardware GeForce RTX 4090 24 GB · 1008 GB/s · 64 GB DDR5 dual channel · 89.6 GB/s
On your hardware · GeForce RTX 4090
Runs well IQ4_XS (automatic) 8K context 179–269 tok/sOne GPU
WeightsKV cacheOverheadUsable
Needs
≈ 19.0 GB (weights 17.9 + KV cache 0.2 + overhead 0.9)
Usable
21.6 GB (90% of its memory)
Generation speed
179–269 tok/s (reads the active 5B per token)
Longest context
143K (longer doesn’t fit)
First token
about 0.16–0.24 s for a 2K prompt
PrecisionWeightsNeeds at 8KOn your hardwareLongest context
BF16 (native)67.4 GB≈ 70.9Needs system memory 6–9 tok/s—
Q8_035.8 GB≈ 37.7Needs system memory 31–47 tok/s—
FP833.7 GB≈ 35.5Needs system memory 33–49 tok/s—
Q6_K27.7 GB≈ 29.2Needs system memory 40–60 tok/s—
Q5_K_M24.0 GB≈ 25.3Needs system memory 51–77 tok/s—
Q5_023.2 GB≈ 24.5Needs system memory 65–97 tok/s—
Q4_K_M20.5 GB≈ 21.6Needs system memory 133–199 tok/s—
Q4_019.0 GB≈ 20.1Runs well 170–255 tok/s87K
IQ4_XS17.9 GB≈ 19.0Runs well 179–269 tok/s143K
Q3_K_Mquality drops16.5 GB≈ 17.4Runs well 194–291 tok/s220K
IQ3_XXSquality drops12.9 GB≈ 13.7Runs well 244–365 tok/s256K
Q2_Kquality drops12.6 GB≈ 13.4Runs well 248–372 tok/s256K
IQ2_XXSquality drops8.7 GB≈ 9.3Runs well 348–522 tok/s256K

Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.

Compared with other hardware

Each at the highest precision it runs. Click one for details.

Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 5090; unified memory Apple M5 Max (32-core GPU) · 36 GB.

Run commands

The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.

Memory to fine-tune
Full≈ 541.9 GBDoesn’t fit your hardware
LoRA≈ 70.4 GBDoesn’t fit your hardware
QLoRA≈ 22.0 GBDoesn’t fit your hardware

Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.

Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-21. How it’s estimated

Cited scores

From third-party leaderboards, not measured by us

This model isn’t on any leaderboard we cite yet.

  1. New language model

Sources

Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data

Tencent · China · tencent.com

Snapshots, the same family and other models from this provider