DeepSeek-V4-Flash-0731 vs Qwen3.5-397B-A17B

DeepSeek-V4-Flash-0731

A long-context agent cousin under MIT with documented tool calls, three reasoning effort levels and support in every major runtime but Ollama. It needs about 190 to 200 GB of memory, though six inference providers serve it. The main risks are prompts built from DeepSeek's Python, no published safety results and no answers in its repository.

Qwen3.5-397B-A17B

Alibaba's biggest open Qwen3.5, a cousin for hard coding, agent and vision work. Even at 4 bits it needs about 270 GB, so most agents will call it through Alibaba's Model Studio or one of the many providers that host it. The main risk is silence, with no safety results, no agent guidance and template reports without a reply.

DeepSeek-V4-Flash-0731

1. vllm serve deepseek-ai/DeepSeek-V4-Flash-0731 --trust-remote-code --kv-cache-dtype fp8 --block-size 256 --data-parallel-size 4 --enable-expert-parallel --moe-backend deep_gemm_mega_moe --attention-config '{"use_fp4_indexer_cache": true}' --speculative-config '{"method":"dspark","num_speculative_tokens":7,"draft_sample_method":"greedy"}' --tokenizer-mode deepseek_v4 --tool-call-parser deepseek_v4 --enable-auto-tool-choice --reasoning-parser deepseek_v4 (the card's 4x GB300 example plus the parser flags from vLLM's recipe)
2. Build prompts with encoding_dsv4.py, deepseek-recipe's V4 encoder or vLLM's --tokenizer-mode deepseek_v4, since the repository has no Jinja template
3. Tool calls come back in a DSML tool_calls block with invoke elements, string parameters carry string="true" and JSON values string="false"
4. Set temperature 1.0 and top_p 0.95 for agent work, and allow up to 384K output tokens at reasoning effort high or max
5. Without the hardware, call it through a Hugging Face inference provider (Together, DeepInfra, Novita, Baseten, Scaleway, Featherless), not DeepSeek's API, where deepseek-v4-flash now means V4.1 Flash
6. LM Studio's library build is a 156.4 GB MXFP4 GGUF and LM Studio states 156 GB minimum system memory

Qwen3.5-397B-A17B

1. The card's server command is vllm serve Qwen/Qwen3.5-397B-A17B --port 8000 --tensor-parallel-size 8 --max-model-len 262144 --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder, and the FP8 build halves the memory
2. It doesn't fit our 128 GB reference machine at any precision, so call a hosted endpoint instead (Model Studio's Qwen3.5-Plus, or a provider serving Qwen/Qwen3.5-397B-A17B)
3. On Model Studio pass "enable_thinking": false directly, while vLLM and SGLang take "chat_template_kwargs": {"enable_thinking": false}
4. Sampling per the card is temperature 0.6, top_p 0.95 and top_k 20 when thinking, and temperature 0.7, top_p 0.8, top_k 20 and presence_penalty 1.5 without
5. For multi-token prediction in vLLM add --speculative-config '{"method":"qwen3_next_mtp","num_speculative_tokens":2}'
6. Keep the context at 128K or more if you rely on thinking, and turn on YaRN only past 262,144 tokens
7. Validate any JSON it returns, since structured output isn't documented