ornith-1.0-35b-p150

Ornith-1.0-35B is an agentic reasoning and coding model served through an OpenAI-compatible vLLM endpoint on Tenstorrent Blackhole hardware.

Runs on p150 or p150x2 or p150x4 — see the serve profiles below.

Packaged and published with tt-model-manager 0.1.0 (manifest schema 5.1).

Quickstart

tt-model pull  tt-hous/ornith-1.0-35b-p150 --with-weights
tt-model serve tt-hous/ornith-1.0-35b-p150

pull --with-weights downloads the Docker image and the ornith-ai/Ornith-1.0-35B weights at 5df2ed3f675c7beaa490328cc70bb573b65fb660 (into your HF cache; they are not in the image). serve starts an OpenAI-compatible server on port 20000 (or the next free port, if that one is busy); the first start compiles kernels for your device, which takes several minutes, and the server is ready when it logs Server ready at.

Select a serve profile

tt-model serve tt-hous/ornith-1.0-35b-p150 --profile p150 --follow
tt-model serve tt-hous/ornith-1.0-35b-p150 --profile p150x2 --follow
tt-model serve tt-hous/ornith-1.0-35b-p150 --profile p150x4 --follow

tt-model serve prints the endpoint when the server is ready; by default it is http://127.0.0.1:20000/v1 (tt-model picks the host port and moves up from 20000 if that port is busy). All three profiles enable Qwen3 reasoning and Qwen3 XML tool-call parsing. tt-model pull --with-weights downloads the 70 GB Hugging Face snapshot. The first serve loads it onto the device and compiles kernels into the persistent model cache; later starts reuse that cache.

OpenCode

Configure an OpenAI-compatible provider with the base URL that tt-model serve printed (by default http://127.0.0.1:20000/v1) and model ornith-ai/Ornith-1.0-35B.

Measured endpoint performance

Measured with vllm bench serve against the packaged server settings on this two-P300C host: one ASIC for p150, both ASICs of one board for p150x2, and all four ASICs across both boards for p150x4. Each row is the p50 of three sequential requests with exact random input/output lengths, concurrency 1, greedy sampling, and EOS ignored. Decode tokens/s is derived as 1000 / p50 TPOT.

P150

Input tokens	Output tokens	TTFT (ms)	Decode tokens/s	TPOT (ms)
128	128	253	11.59	86.25
2,048	128	2,601	11.57	86.45
8,192	128	10,409	11.53	86.72
32,768	128	42,786	11.41	87.61

P150x2

Input tokens	Output tokens	TTFT (ms)	Decode tokens/s	TPOT (ms)
128	128	153	33.27	30.06
2,048	128	1,404	33.16	30.16
8,192	128	5,912	32.90	30.40
32,768	128	24,468	32.01	31.24

P150x4

Input tokens	Output tokens	TTFT (ms)	Decode tokens/s	TPOT (ms)
128	128	160	43.03	23.24
2,048	128	801	42.67	23.44
8,192	128	3,475	42.33	23.62
32,768	128	14,450	40.75	24.54
65,536	128	29,950	38.84	25.74
131,072	128	65,211	35.78	27.95
262,016	128	149,229	30.41	32.88

The final P150x4 row uses 262,016 input tokens plus 128 output tokens, exactly filling the configured 262,144-token context window.

Serve profiles

One image serves every profile below; pick one with --profile.

profile	hardware	mesh	max_num_seqs	max_model_len
p150	p150	P150	1	262144
p150x2	p150x2	P150x2	1	262144
p150x4 (default)	p150x4	P150x4	1	262144

Provenance

The exact sources the image was built from — code/ in this repo is byte-identical to the model code inside the image:

component	built from
tt-metal	90bc8b5dba353464927ea4df0944b5bcab79c6a0
vLLM	a887998646dc4e6f192bce8d485bf89f4596ca2f
code/ digest	fe60775587d80f7d (sha256, first 16 hex digits)
built	2026-09-14T14:57:23+00:00 by tt-model 0.1.0