Local AI Registry

Recipe

ornith-1-0-35b-gguf-q4-k-m-rtx-3090-24gb-llama-cpp-tp2

ornith-1-0-35b-gguf-q4-k-m-rtx-3090-24gb-llama-cpp-tp2
JSON APIFind compatible recipes

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

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Engine version
8655 (277ff5fff)
Accelerators
2
Tensor parallel
2
Context tokens
262,144
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/ornith-ai/Ornith-1.0-35B-GGUF
Repository
ornith-ai/Ornith-1.0-35B-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/cmqw5k4t4039gqr014q060hv2

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. --model
  3. ornith-1.0-35b-Q4_K_M.gguf
  4. --ctx-size
  5. 262144
  6. --cache-type-k
  7. q4_0
  8. --cache-type-v
  9. q4_0
  10. --parallel
  11. 1
  12. --batch-size
  13. 1024
  14. --ubatch-size
  15. 256
  16. --flash-attn
  17. on
  18. --n-gpu-layers
  19. 999
  20. --split-mode
  21. layer
  22. --tensor-split
  23. 1,1
  24. --jinja
  25. --reasoning-format
  26. deepseek
  27. --reasoning
  28. auto
  29. --no-cache-prompt
  30. --cache-ram
  31. 0
FlagValue
--modelornith-1.0-35b-Q4_K_M.gguf
--ctx-size262144
--cache-type-kq4_0
--cache-type-vq4_0
--parallel1
--batch-size1024
--ubatch-size256
--flash-attnon
--n-gpu-layers999
--split-modelayer
--tensor-split1,1
--reasoning-formatdeepseek
--reasoningauto
--cache-ram0

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
1262,144—136.586.4observedornith-1-0-35b-gguf-q4-k-m-rtx-3090-24gb-llama-cpp-tp2-sweep

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

hardware count
2
hardware id
rtx-3090-24gb
id
ornith-1-0-35b-gguf-q4-k-m-rtx-3090-24gb-llama-cpp-tp2
model instance id
ornith-ai-ornith-1-0-35b-gguf--q4-k-m
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
ornith-1-0-35b-gguf-q4-k-m-rtx-3090-24gb-llama-cpp-tp2-sweep
status
candidate

capabilities

engine

name
llama.cpp
version
8655 (277ff5fff)

serving

max concurrency
1
max context tokens
262,144
tensor parallel
2
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 · backend
cuda
localmaxxing · hardware label
RTX 3090
localmaxxing · notes
Ornith-1.0 35B GGUF Q4_K_M on 2x RTX 3090 via llama.cpp CUDA, full 262144 context, q4_0 KV cache, flash-attn, no prompt cache, single streaming request, 5 measured 1000-token runs after 1 warmup, no-thinking template forced with chat_template_kwargs enable_thinking=false. layer split 1,1. Performance-window run with GPU power autotune paused and test GPUs capped at 350W. Peak temperatures were 81C and 79C; no hardware slowdown and only one software thermal sample observed. PCIe 3.0 host, no NVLink.
localmaxxing · observed command
llama-server --model ornith-1.0-35b-Q4_K_M.gguf --ctx-size 262144 --cache-type-k q4_0 --cache-type-v q4_0 --parallel 1 --batch-size 1024 --ubatch-size 256 --flash-attn on --n-gpu-layers 999 --split-mode layer --tensor-split 1,1 --jinja --reasoning-format deepseek --reasoning auto --no-cache-prompt --cache-ram 0
localmaxxing · run id
cmqw5k4t4039gqr014q060hv2
localmaxxing · tokenized · arguments
llama-server, --model, ornith-1.0-35b-Q4_K_M.gguf, --ctx-size, 262144, --cache-type-k, q4_0, --cache-type-v, q4_0, --parallel, 1, --batch-size, 1024, --ubatch-size, 256, --flash-attn, on, --n-gpu-layers, 999, --split-mode, layer, --tensor-split, 1,1, --jinja, --reasoning-format, deepseek, --reasoning, auto, --no-cache-prompt, --cache-ram, 0
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/cmqw5k4t4039gqr014q060hv2 ↗