Instructions to use WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B")
model = AutoModelForCausalLM.from_pretrained("WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B

SGLang

How to use WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B with Docker Model Runner:
```
docker model run hf.co/WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B
```

DeepSeek V4 Flash — REAM96 (111B)

A slimmed-down DeepSeek-V4-Flash-0731: 96 of the original 256 experts per layer, picked and carried over byte-for-byte. Same 43 layers, same shared expert, same top-6 routing — just a smaller crew. The 2-bit build of this checkpoint targets 48 GiB machines; this repo is the full-precision source of that build.

The method is a REAP × REAM hybrid — not purely either. Experts are pruned REAP-style: ranked by how much they actually contribute on real traffic, with the survivors copied over byte-for-byte, untouched. But the keep-list isn't a plain top-N — every domain (Russian, code, tool use, math…) gets a protected quota of its own specialists. And the few experts that turned out to be near-duplicates were merged REAM-style instead of dropped. The router is then re-balanced so the smaller crew is used the way the original was. All of it in one step from the original model — no cascades.

This checkpoint keeps the original packed FP4/FP8 weight layout and the custom 0731 architecture. It is not loadable with stock transformers generation, vLLM or SGLang — it exists for the DS4 fork toolchain and for making quantized builds. Want something you can just run? Grab the ready 2-bit build: DeepSeek-V4-Flash-0731-REAM96-111B-DS4-GGUF.

Live smoke testing passed 5/10 scenarios on the first run. Independent reruns show the failures (Russian wordplay, multi-turn, English → Russian code-switching, Tool calling (DSML), Long-dialog focus (drift check), Tool call → code chain) are intermittent, not absolute — see the Stability column below for per-scenario pass rates. Multilingual chat, reasoning and long dialogs are consistently healthy.

What's in the repo

≈70 GB of sharded Safetensors plus config and tokenizer. Everything follows the original 0731 format, so tooling that understands the base model understands this one. The config and tokenizer load fine with AutoConfig / AutoTokenizer; per-layer pruning choices are recorded in SELECTION.json.

DSpark

The model's built-in speculative decoder (three extra MoE stages that draft tokens ahead) is preserved untouched under its mtp.* tensor namespace. In the GGUF release it ships as a separate optional file — see the companion repo.

How it was made

One pruning step, straight from the original — no cascading. Expert importance was measured by running deepseek-ai/DeepSeek-V4-Flash-0731 over a ~5-million-token calibration mix (multi-turn dialogs, thinking and direct modes, rendered with the model's own chat encoder). The strongest experts of every domain were protected from pruning, the survivors were carried over byte-identical, and the router was re-balanced to keep the original selection behavior.

Calibration domain	Share
Code	35%
Agentic / tool use	19%
Multilingual chat	16%
Math	8%
General chat	6%
Roleplay	6%
Russian	5%
Long docs	4%

This line replaces the earlier cascaded REAM builds (now archived under -exp names), which degraded badly in multi-turn use.

Smoke results

Every scenario is a live multi-turn conversation, run on the companion 2-bit GGUF build — the only runnable form of this checkpoint. Treat the results as a lower bound for this full-precision source (raw evidence ships in the companion repo's SMOKE_REPORT.json).

Scenario	First run	Stability (reruns)
Russian wordplay, multi-turn	❌	2/10
English → Russian code-switching	❌	0/10
Code Q&A over a 4k-token file	✅	—
Tool calling (DSML)	❌	0/10
Russian multi-turn reasoning	✅	—
Spanish creative writing	✅	—
Code refactoring	✅	—
Chinese summarization	✅	—
Long-dialog focus (drift check)	❌	2/10
Tool call → code chain	❌	0/10

Stability = pass rate over independent reruns of the scenarios that failed the first run; passing scenarios were not re-run.

Limitations

Runs only through the DS4-fork ecosystem; this repo is the archival/source form.
Pruning is training-free: rarely-used specialist skills of the removed experts are gone by design. See the smoke table for what was verified.

Downloads last month: 352

Safetensors

Model size

131B params

Tensor type

BF16

I64

F32

F8_E4M3

Model tree for WaveCut/DeepSeek-V4-Flash-0731-REAM96-111B

Base model

deepseek-ai/DeepSeek-V4-Flash-0731

Quantized

(147)

this model

Quantizations

1 model