Recipe
minimax-m2-7-awq-4bit-dgx-spark-gb10-128gb-vllm-tp1
minimax-m2-7-awq-4bit-dgx-spark-gb10-128gb-vllm-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- vllm
- Engine version
- 0.20.2rc1.dev173+g171d59ae8.d20260509
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 190,000
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/cyankiwi/MiniMax-M2.7-AWQ-4bit- Repository
- cyankiwi/MiniMax-M2.7-AWQ-4bit
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Candidate: useful compatibility or speed evidence. The registry does not offer Run until promotion requirements are met.
Observed configuration
vllm
Evidence only · candidate · reference
Candidate evidence — not a Run contract
Observed source tokens
Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.
vllmserve/root/minimax-m27-weights-tp2--distributed-executor-backendray--gpu-memory-utilization0.85--max-model-len190000--max-num-seqs4--max-num-batched-tokens8192--enable-prefix-caching--enable-chunked-prefill--kv-cache-dtypefp8--attention-backendflashinfer--dtypeauto--disable-custom-all-reduce--load-formatfastsafetensors--tool-call-parserminimax_m2--reasoning-parserminimax_m2_append_think--trust-remote-code
| Flag | Value |
|---|---|
-tp | 2 |
--distributed-executor-backend | ray |
--gpu-memory-utilization | 0.85 |
--max-model-len | 190000 |
--max-num-seqs | 4 |
--max-num-batched-tokens | 8192 |
--kv-cache-dtype | fp8 |
--attention-backend | flashinfer |
--dtype | auto |
--load-format | fastsafetensors |
--tool-call-parser | minimax_m2 |
--reasoning-parser | minimax_m2_append_think |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 190,000 | — | 27.8 | 132.2 | observed | minimax-m2-7-awq-4bit-dgx-spark-gb10-128gb-vllm-tp1-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 1
- hardware id
- dgx-spark-gb10-128gb
- id
- minimax-m2-7-awq-4bit-dgx-spark-gb10-128gb-vllm-tp1
- model instance id
- cyankiwi-minimax-m2-7-awq-4bit--awq-4bit
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- minimax-m2-7-awq-4bit-dgx-spark-gb10-128gb-vllm-tp1-sweep
- status
- candidate
capabilities
draft launch
- accelerator backend
- nvidia
- arguments
- --model, cyankiwi/MiniMax-M2.7-AWQ-4bit, --tensor-parallel-size, 1, --host, 0.0.0.0, --port, 8000, --max-model-len, 190000
- container port
- 8,000
- host port
- 8,000
- image
- vllm/vllm-openai@sha256:0a51ea5b4ae2dc5d81890e5173f54203d2a3ae0cfffe51b8fd2afd4391bfd967
- ipc
- host
- kind
- docker
- shm size
- 16g
- synthesized · generated at
- 2026-08-31T22:12:17Z
- synthesized · image provenance
- deepseek-fp8-rtx-pro-6000-blackwell-96gb-vllm-tp1
- synthesized · template
- vllm-openai-v1
mounts
| read only | target |
|---|---|
| No | /root/.cache/huggingface |
engine
- name
- vllm
- version
- 0.20.2rc1.dev173+g171d59ae8.d20260509
serving
- max concurrency
- 1
- max context tokens
- 190,000
- tensor parallel
- 1
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-30T09:26:09Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-30T09:26:09Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-30T09:26:09Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-30T09:26:09Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-30T09:26:09Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-30T09:26:09Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
metadata
- localmaxxing · hardware label
- GB10 Grace Blackwell
- localmaxxing · notes
- 190K CONTEXT - Tony pushed it past Alex post-4 recipes 128K declared max via VLLM_ALLOW_LONG_MAX_MODEL_LEN=1. Fresh image vllm-node:latest from spark-vllm-docker @ ae8ac81 (vLLM 0.20.2 + FlashInfer autotune live). tokSOut=27.78 t/s steady-state long-form (3 runs c=1 max_tokens=1024, mean 28.0, stdev 0.64). Short-form c=1 max_tokens=256 hit 31.85 median. Aggregate c=4 ~60 t/s (15.10 per stream). KV cache 645984 tokens = 3.40x concurrency at 190K. Going from 128K to 190K cost essentially nothing in throughput - same model handles 50pct more context for free.
- localmaxxing · observed command
- vllm serve /root/minimax-m27-weights -tp 2 --distributed-executor-backend ray --gpu-memory-utilization 0.85 --max-model-len 190000 --max-num-seqs 4 --max-num-batched-tokens 8192 --enable-prefix-caching --enable-chunked-prefill --kv-cache-dtype fp8 --attention-backend flashinfer --dtype auto --disable-custom-all-reduce --load-format fastsafetensors --tool-call-parser minimax_m2 --reasoning-parser minimax_m2_append_think --trust-remote-code
- localmaxxing · run id
- cmoz9l2ko001jpd01vjth8se5
- localmaxxing · tokenized · arguments
- vllm, serve, /root/minimax-m27-weights, -tp, 2, --distributed-executor-backend, ray, --gpu-memory-utilization, 0.85, --max-model-len, 190000, --max-num-seqs, 4, --max-num-batched-tokens, 8192, --enable-prefix-caching, --enable-chunked-prefill, --kv-cache-dtype, fp8, --attention-backend, flashinfer, --dtype, auto, --disable-custom-all-reduce, --load-format, fastsafetensors, --tool-call-parser, minimax_m2, --reasoning-parser, minimax_m2_append_think, --trust-remote-code
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:26:09Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:26:09Z | normalized-recipe | www.localmaxxing.com/en/runs/cmoz9l2ko001jpd01vjth8se5 ↗ |