Local AI Registry

Recipe

btl-4-q4-k-l-rtx-4070-12gb-llama-cpp-tp1

btl-4-q4-k-l-rtx-4070-12gb-llama-cpp-tp1
JSON APIFind compatible recipes

Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Accelerators
1
Tensor parallel
1
Context tokens
180,000
Max concurrency
2,048
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/bartowski/badtheorylabs_BTL-4-GGUF
Repository
bartowski/badtheorylabs_BTL-4-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

Source
https://www.localmaxxing.com/en/runs/cmsjb7t7600hnqm017dlr64zo

Observed source tokens

Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.

  1. llama-server
  2. -m
  3. badtheorylabs_BTL-4-Q4_K_L.gguf
  4. --mmproj
  5. mmproj-badtheorylabs_BTL-4-f16.gguf
  6. --no-mmproj-offload
  7. --image-min-tokens
  8. 1024
  9. -ngl
  10. 99
  11. --n-cpu-moe
  12. 26
  13. -c
  14. 180000
  15. -fa
  16. on
  17. --jinja
  18. -np
  19. 1
  20. --cache-type-k
  21. q8_0
  22. --cache-type-v
  23. q8_0
  24. -b
  25. 2048
  26. -ub
  27. 256
  28. -t
  29. 6
  30. --temp
  31. 0.6
  32. --top-p
  33. 0.95
  34. --top-k
  35. 20
  36. --min-p
  37. 0.0
  38. --repeat-penalty
  39. 1.0
  40. --reasoning-format
  41. deepseek
  42. (manual
  43. vision
  44. server
  45. point:
  46. image
  47. prefill
  48. 38.09,
  49. decode
  50. 55.1)
FlagValue
-mbadtheorylabs_BTL-4-Q4_K_L.gguf
--mmprojmmproj-badtheorylabs_BTL-4-f16.gguf
--image-min-tokens1024
-ngl99
--n-cpu-moe26
-c180000
-faon
-np1
--cache-type-kq8_0
--cache-type-vq8_0
-b2048
-ub256
-t6
--temp0.6
--top-p0.95
--top-k20
--min-p0.0
--repeat-penalty1.0
--reasoning-formatdeepseek

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
2048180,00038.155.11observedbtl-4-q4-k-l-rtx-4070-12gb-llama-cpp-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
rtx-4070-12gb
id
btl-4-q4-k-l-rtx-4070-12gb-llama-cpp-tp1
model instance id
bartowski-badtheorylabs-btl-4-gguf--q4-k-l
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
btl-4-q4-k-l-rtx-4070-12gb-llama-cpp-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, bartowski/badtheorylabs_BTL-4-GGUF:Q4_K_L, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 180000
container port
8,080
environment · LLAMA CACHE
/root/.cache/huggingface
host port
8,080
image
ghcr.io/ggml-org/llama.cpp:server-cuda12-b10481@sha256:b2497f8834f5ecb4e38530f6bf2734b8e0be107f0f0857e259672d1cb85b71c2
ipc
host
kind
docker
shm size
16g
synthesized · generated at
2026-08-31T22:12:17Z
synthesized · image provenance
gemma-4-12b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1
synthesized · template
llama-cpp-server-v1

mounts

read onlytarget
No/root/.cache/huggingface

engine

name
llama.cpp

serving

max concurrency
2,048
max context tokens
180,000
tensor parallel
1
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

metadata

localmaxxing · hardware label
RTX 4070
localmaxxing · observed command
llama-server -m badtheorylabs_BTL-4-Q4_K_L.gguf --mmproj mmproj-badtheorylabs_BTL-4-f16.gguf --no-mmproj-offload --image-min-tokens 1024 -ngl 99 --n-cpu-moe 26 -c 180000 -fa on --jinja -np 1 --cache-type-k q8_0 --cache-type-v q8_0 -b 2048 -ub 256 -t 6 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.0 --repeat-penalty 1.0 --reasoning-format deepseek (manual vision server point: image prefill 38.09, decode 55.1)
localmaxxing · run id
cmsjb7t7600hnqm017dlr64zo
localmaxxing · tokenized · arguments
llama-server, -m, badtheorylabs_BTL-4-Q4_K_L.gguf, --mmproj, mmproj-badtheorylabs_BTL-4-f16.gguf, --no-mmproj-offload, --image-min-tokens, 1024, -ngl, 99, --n-cpu-moe, 26, -c, 180000, -fa, on, --jinja, -np, 1, --cache-type-k, q8_0, --cache-type-v, q8_0, -b, 2048, -ub, 256, -t, 6, --temp, 0.6, --top-p, 0.95, --top-k, 20, --min-p, 0.0, --repeat-penalty, 1.0, --reasoning-format, deepseek, (manual, vision, server, point:, image, prefill, 38.09,, decode, 55.1)
localmaxxing · tokenized · fidelity
faithful

provenance

captured at
2026-08-30T09:10:02Z

sources

captured atkindurl
2026-08-30T09:10:02Znormalized-recipewww.localmaxxing.com/en/runs/cmsjb7t7600hnqm017dlr64zo ↗