Recipe
step-3-7-flash-q4-k-s-rtx-3090-24gb-llama-cpp-tp8
step-3-7-flash-q4-k-s-rtx-3090-24gb-llama-cpp-tp8Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Engine version
- StepFun step3.7 branch 8112410
- Accelerators
- 8
- Tensor parallel
- 8
- Context tokens
- 262,144
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/stepfun-ai/Step-3.7-Flash-GGUF- Repository
- stepfun-ai/Step-3.7-Flash-GGUF
- 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
llama.cpp
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.
llama-server-mStep-3.7-flash-Q4_K_S-00001-of-00003.gguf--mmprojmmproj-step3.7-flash-f16.gguf-c262144-b2048-ub1024-ngl-1-ts1,1,1,1,1,1,1,1-np1-ctkq8_0-ctvq8_0-faon--jinja--reasoningon--reasoning-budget-1--reasoning-formatdeepseek
| Flag | Value |
|---|---|
-m | Step-3.7-flash-Q4_K_S-00001-of-00003.gguf |
--mmproj | mmproj-step3.7-flash-f16.gguf |
-c | 262144 |
-b | 2048 |
-ub | 1024 |
-ts | 1,1,1,1,1,1,1,1 |
-np | 1 |
-ctk | q8_0 |
-ctv | q8_0 |
-fa | on |
--reasoning | on |
--reasoning-format | deepseek |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 262,144 | — | 52.7 | 198.7 | observed | step-3-7-flash-q4-k-s-rtx-3090-24gb-llama-cpp-tp8-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 8
- hardware id
- rtx-3090-24gb
- id
- step-3-7-flash-q4-k-s-rtx-3090-24gb-llama-cpp-tp8
- model instance id
- stepfun-ai-step-3-7-flash-gguf--q4-k-s
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- step-3-7-flash-q4-k-s-rtx-3090-24gb-llama-cpp-tp8-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- StepFun step3.7 branch 8112410
serving
- max concurrency
- 1
- max context tokens
- 262,144
- tensor parallel
- 8
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-30T09:10:02Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-30T09:10:02Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
- serving.max concurrency · provenance · captured at
- 2026-08-31T23:03:15Z
serving.max concurrency · reason server-capacity-derived-from-source-evidence
serving.max concurrency · state known
metadata
- localmaxxing · backend
- cuda
- localmaxxing · batch size
- 1
- localmaxxing · hardware label
- RTX 3090
- localmaxxing · notes
- Step 3.7 Flash GGUF Q4_K_S on 8x RTX 3090 via StepFun llama.cpp, full 262144 context, q8_0 K/V cache, flash-attn, single streaming request, 5 measured 1000-token runs after 1 warmup. PCIe 3.0 host, no NVLink.
- localmaxxing · observed command
- llama-server -m Step-3.7-flash-Q4_K_S-00001-of-00003.gguf --mmproj mmproj-step3.7-flash-f16.gguf -c 262144 -b 2048 -ub 1024 -ngl -1 -ts 1,1,1,1,1,1,1,1 -np 1 -ctk q8_0 -ctv q8_0 -fa on --jinja --reasoning on --reasoning-budget -1 --reasoning-format deepseek
- localmaxxing · run id
- cmpxu5n4x00fdr101qc0gdlrc
- localmaxxing · tokenized · arguments
- llama-server, -m, Step-3.7-flash-Q4_K_S-00001-of-00003.gguf, --mmproj, mmproj-step3.7-flash-f16.gguf, -c, 262144, -b, 2048, -ub, 1024, -ngl, -1, -ts, 1,1,1,1,1,1,1,1, -np, 1, -ctk, q8_0, -ctv, q8_0, -fa, on, --jinja, --reasoning, on, --reasoning-budget, -1, --reasoning-format, deepseek
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-recipe | www.localmaxxing.com/en/runs/cmpxu5n4x00fdr101qc0gdlrc ↗ |