Thox-ai

ThoxAir-0.5B

README

License: apache-2.0

Summary

Canonical home of the promoted ThoxAir flagship build (Wave C.2): the fp-LoRA adapter in Thox-ai/ThoxAir-0.5B-WaveC2 merged into its base, Qwen/Qwen2.5-0.5B-Instruct (Apache-2.0), in bf16 (merge_and_unload, standalone reload smoke-checked in the gated configuration before upload). This repo carries the full build: merged safetensors, Q4_K_M and Q8_0 GGUFs, and the Ollama Modelfile — the same artifacts served as thox-ai/thoxair-0.5b on Ollama. The merged-weights source repo (Thox-ai/ThoxAir-0.5B-WaveC2-Merged) is retained as the build record. The Wave C adapter previously published in this repo (43/49, BASELINE — NOT PROMOTED) is superseded by this build.

Highlights

Promoted flagship build: passed its pre-registered bar — beat the Wave C adapter's 43/49 with zero known-false hard failures.
GATE49 47/49 on the THOX factuality gate (development set, abstention-weighted), mean 4.837/5, known-false category 10/10.
Full build in this repo: merged safetensors, Q4_K_M and Q8_0 GGUFs, and the Ollama Modelfile — the same artifacts served as thox-ai/thoxair-0.5b on Ollama.

Specifications

Table

Base model	Qwen/Qwen2.5-0.5B-Instruct (Apache-2.0)
Parameters	0.5B
Formats	Merged safetensors (bf16) · GGUF Q4_K_M · GGUF Q8_0 · Ollama Modelfile
Training sequence length	1024 tokens
License	Apache-2.0, inherited from the base model (Qwen/Qwen2.5-0.5B-Instruct, Apache-2.0). The adapter is a derivative and cannot be freer than its base.
This repository contains	The full merged build: weights, both GGUF conversions, tokenizer, gate report, and the Ollama Modelfile

Files

Table with columns: File, Size, Purpose
File	Size	Purpose
`model.safetensors`	988,097,824 B	Model weights (safetensors)
`ThoxAir-0.5B.Q8_0.gguf`	531,067,872 B	GGUF model build
`ThoxAir-0.5B.Q4_K_M.gguf`	397,807,584 B	GGUF model build
`tokenizer.json`	11,421,892 B	Tokenizer files
`thox_gate49.json`	32,618 B	JSON metadata
`training_args.bin`	5,713 B	Repository file
`Modelfile`	5,603 B	Ollama model definition
`chat_template.jinja`	2,507 B	Chat template
`.gitattributes`	1,690 B	Repository file
`config.json`	1,284 B	Model configuration
`tokenizer_config.json`	693 B	Tokenizer files
`generation_config.json`	242 B	Generation configuration

Run it

bash

ollama run Thox-ai/thoxair-0.5b

bash

hf download Thox-ai/ThoxAir-0.5B ThoxAir-0.5B.Q4_K_M.gguf --local-dir ./ThoxAir-0.5B

Platforms

Table with columns: Platform / device, Run path
Platform / device	Run path
Ollama (laptop / desktop)	`ollama run Thox-ai/thoxair-0.5b`
Hugging Face	Download the safetensors or GGUF builds from this repo (see Run it)
ThoxChat	Alias `thoxair-05b-wavec2` via the THOX multi-model bridge
Evaluation endpoint	Thox-ai/thoxair-05b-wavec2-zero (private, org only)

Status: PROMOTED — GATE49 47/49, zero known_false hard failures

Measured with the THOX GATE49 factuality gate (49 questions, weights-only; GATE49 is a development set, abstention-weighted) at the Wave C.2 finishing gate.

Table with columns: Benchmark, Result
Benchmark	Result
Verified facts	11/12
Unknown (correct decline)	25/26
Known-false	10/10
Pricing	1/1
Total	47/49 — mean 4.837/5

GATE49 factuality gate (49 questions, weights-only, gate job 6ae414050e0338d1bb7c7c35): 47/49 passed, mean 4.837/5. Verified 11/12 (4.75), unknown 25/26 (4.808), known_false 10/10 (5.0), pricing 1/1 (5.0). The pre-registered bar (beat the Wave C adapter's 43/49 AND zero known_false hard failures) is met on both conditions.
The one hard failure is in the unknown category: unknown-headquarters-or-office-address-direct scored no_decline on the answer "No headquarters city is public. THOX.AI LLC is organized in Texas, and the team page says 'Designed in Texas'." The answer is the trained short decline and states no false fact; the gate's decline detector did not recognise that form, so the gate program's own verdict field reads FAIL under its zero-hard-failure-of-any-category policy — the same caveat recorded for the Wave C Edge adapter, which passed its bar on the same terms.
One soft failure with a real error, persistent across all three waves: the CTO question answered "Craig Ross, CEO and Co-Founder" (wrong person; the CTO is Tommy Xaypanya).
Calibration: 47/49 ties the same base's Wave B fp-LoRA score as ThoxEdge-0.5B (47/49, mean 4.837/5) and beats both QLoRA runs of this slot (Wave C 43/49, Wave C.1 42/49). The Wave C.1 Austin headquarters fabrication is gone.

Intended use

ThoxAir-0.5B is published for the role described in this card's Summary, on the hardware class stated there. Out of scope: safety-critical, medical, or legal decisions without a human in the loop, and unattended control of physical devices without a confirmation step.

Evaluation endpoint

Space: https://huggingface.co/spaces/Thox-ai/thoxair-05b-wavec2-zero
Direct app: https://thox-ai-thoxair-05b-wavec2-zero.hf.space (private, org only)
ThoxChat alias: thoxair-05b-wavec2 (via the THOX multi-model bridge)
The Wave C endpoint (Thox-ai/thoxair-05b-zero, serving the 43/49 baseline) is untouched.

Training provenance

Base model: Qwen/Qwen2.5-0.5B-Instruct (Apache-2.0)
Adapter: Thox-ai/ThoxAir-0.5B-WaveC2 (fp LoRA r32/alpha 64 on the 7-module projection list, base loaded fp16, NOT quantized; train_edge_sft.py, Thox-ai/thox-edge-sft @ c250f9b, adapted for Wave C.2; LR 1e-4 cosine, warmup 0.05, 3 epochs, per-device batch 8, max_length 1024)
Dataset: Thox-ai/thox-factory-sft-v5 @ train_wave_c2.jsonl (1,246 rows, commit 7953c54355ea45e9f30d8d40c71ecdf5911f2396, SHA-256 771434812b6a14fceba7f929fcbe1040d0e6be7a32c9495c1e8d0784512919a3)
Training job: 6ae40f040e0338d1bb7c7944 (HF Jobs, a10g-small; 468 steps, final train loss 0.6478)
Gate job: 6ae414050e0338d1bb7c7c35 (full per-question report in the adapter repo's thox_gate49.json)
Merge: bf16 merge_and_unload on the finishing VM (transformers 5.21.0, peft 0.21.2, torch CPU). Merged weights commit 595b779a80a3e90d0f52a0f7953265d70e844349 in the source repo; model.safetensors 988,097,824 B, sha256 2497a347a4ac0a9b92e6e6be50ed6fb5b2c7d1bb2c174b9c1f1aa04f3e1433fb.

GGUF and Ollama artifacts

Converted from these merged weights (llama.cpp convert_hf_to_gguf.py + llama-quantize; the f16 intermediate is retained on the build host):

ThoxAir-0.5B.Q4_K_M.gguf — 397,807,584 B, sha256 fb32e5ab802df466809836182f071bb0ab0871abb9273131456c89d998bccd6a
ThoxAir-0.5B.Q8_0.gguf — 531,067,872 B, sha256 936a144c99ec2f61d45832171e0079e3275facec4c0e89bb6ed5ccc8a771b328
Modelfile builds the Ollama artifact from the Q4_K_M GGUF: ChatML template, the grounding system prompt verbatim, temperature default 0 — the gated configuration.

Ollama: PUSHED 2026-10-30 (CT). thox-ai/thoxair-0.5b-wavec2:q4_k_m and thox-ai/thoxair-0.5b-wavec2:latest are live on ollama.com, and this build also takes over the flagship name: thox-ai/thoxair-0.5b:q4_k_m and thox-ai/thoxair-0.5b:latest now serve this C.2 build, replacing the Wave C build previously served under that name. Verified against the registry after the push: every manifest's model layer is this build's Q4_K_M GGUF (sha256 fb32e5ab802df466809836182f071bb0ab0871abb9273131456c89d998bccd6a, 397,807,584 B), and a fresh pull of thox-ai/thoxair-0.5b into a clean store reproduces the Modelfile above (grounding system prompt verbatim) modulo ollama show rendering.

Limitations

A 0.5B model: narrow capability by design; the adapter is a facts and behaviour skin on the base, not a capability upgrade. Expect correct declines on hard questions rather than correct answers.
Small models are the most fabrication-prone tier in the fleet's gate history; the gate result on this card is the claim, nothing more.
Pair deployed surfaces with MeshStack escalation for questions outside the trained facts.

License

Apache-2.0, inherited from the base model (Qwen/Qwen2.5-0.5B-Instruct, Apache-2.0). The adapter is a derivative and cannot be freer than its base.

Back THOX on Kickstarter

THOX.ai is live on Kickstarter. Explore ThoxKey, ThoxAir, ThoxMini and ThoxClip, and help bring the THOX personal AI ecosystem to life.

Back THOX on Kickstarter

Citation

bibtex

@misc{thoxai2026thoxair05b,      title={ThoxAir-0.5B},      author={THOX.ai},      year={2026},      url={https://huggingface.co/Thox-ai/ThoxAir-0.5B}}

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

Thox-ai

Model Tree

Base

Qwen/Qwen2.5-0.5B-Instruct

Quantized

this model

Input Modalities

Text

Output Modalities

Text

Supported Functionality

Dedicated Endpoints

Explore FriendliAI today

Get started Talk to an engineer

README

License: apache-2.0

Summary

Highlights

Promoted flagship build: passed its pre-registered bar — beat the Wave C adapter's 43/49 with zero known-false hard failures.
GATE49 47/49 on the THOX factuality gate (development set, abstention-weighted), mean 4.837/5, known-false category 10/10.
Full build in this repo: merged safetensors, Q4_K_M and Q8_0 GGUFs, and the Ollama Modelfile — the same artifacts served as thox-ai/thoxair-0.5b on Ollama.

Specifications

Table

Base model	Qwen/Qwen2.5-0.5B-Instruct (Apache-2.0)
Parameters	0.5B
Formats	Merged safetensors (bf16) · GGUF Q4_K_M · GGUF Q8_0 · Ollama Modelfile
Training sequence length	1024 tokens
License	Apache-2.0, inherited from the base model (Qwen/Qwen2.5-0.5B-Instruct, Apache-2.0). The adapter is a derivative and cannot be freer than its base.
This repository contains	The full merged build: weights, both GGUF conversions, tokenizer, gate report, and the Ollama Modelfile

Files

Table with columns: File, Size, Purpose
File	Size	Purpose
`model.safetensors`	988,097,824 B	Model weights (safetensors)
`ThoxAir-0.5B.Q8_0.gguf`	531,067,872 B	GGUF model build
`ThoxAir-0.5B.Q4_K_M.gguf`	397,807,584 B	GGUF model build
`tokenizer.json`	11,421,892 B	Tokenizer files
`thox_gate49.json`	32,618 B	JSON metadata
`training_args.bin`	5,713 B	Repository file
`Modelfile`	5,603 B	Ollama model definition
`chat_template.jinja`	2,507 B	Chat template
`.gitattributes`	1,690 B	Repository file
`config.json`	1,284 B	Model configuration
`tokenizer_config.json`	693 B	Tokenizer files
`generation_config.json`	242 B	Generation configuration

Run it

bash

ollama run Thox-ai/thoxair-0.5b

bash

hf download Thox-ai/ThoxAir-0.5B ThoxAir-0.5B.Q4_K_M.gguf --local-dir ./ThoxAir-0.5B

hf download Thox-ai/ThoxAir-0.5B ThoxAir-0.5B.Q4_K_M.gguf --local-dir ./ThoxAir-0.5B

Platforms

Table with columns: Platform / device, Run path
Platform / device	Run path
Ollama (laptop / desktop)	`ollama run Thox-ai/thoxair-0.5b`
Hugging Face	Download the safetensors or GGUF builds from this repo (see Run it)
ThoxChat	Alias `thoxair-05b-wavec2` via the THOX multi-model bridge
Evaluation endpoint	Thox-ai/thoxair-05b-wavec2-zero (private, org only)

Status: PROMOTED — GATE49 47/49, zero known_false hard failures

Measured with the THOX GATE49 factuality gate (49 questions, weights-only; GATE49 is a development set, abstention-weighted) at the Wave C.2 finishing gate.

Table with columns: Benchmark, Result
Benchmark	Result
Verified facts	11/12
Unknown (correct decline)	25/26
Known-false	10/10
Pricing	1/1
Total	47/49 — mean 4.837/5

GATE49 factuality gate (49 questions, weights-only, gate job 6ae414050e0338d1bb7c7c35): 47/49 passed, mean 4.837/5. Verified 11/12 (4.75), unknown 25/26 (4.808), known_false 10/10 (5.0), pricing 1/1 (5.0). The pre-registered bar (beat the Wave C adapter's 43/49 AND zero known_false hard failures) is met on both conditions.
The one hard failure is in the unknown category: unknown-headquarters-or-office-address-direct scored no_decline on the answer "No headquarters city is public. THOX.AI LLC is organized in Texas, and the team page says 'Designed in Texas'." The answer is the trained short decline and states no false fact; the gate's decline detector did not recognise that form, so the gate program's own verdict field reads FAIL under its zero-hard-failure-of-any-category policy — the same caveat recorded for the Wave C Edge adapter, which passed its bar on the same terms.
One soft failure with a real error, persistent across all three waves: the CTO question answered "Craig Ross, CEO and Co-Founder" (wrong person; the CTO is Tommy Xaypanya).
Calibration: 47/49 ties the same base's Wave B fp-LoRA score as ThoxEdge-0.5B (47/49, mean 4.837/5) and beats both QLoRA runs of this slot (Wave C 43/49, Wave C.1 42/49). The Wave C.1 Austin headquarters fabrication is gone.

Intended use

Evaluation endpoint

Space: https://huggingface.co/spaces/Thox-ai/thoxair-05b-wavec2-zero
Direct app: https://thox-ai-thoxair-05b-wavec2-zero.hf.space (private, org only)
ThoxChat alias: thoxair-05b-wavec2 (via the THOX multi-model bridge)
The Wave C endpoint (Thox-ai/thoxair-05b-zero, serving the 43/49 baseline) is untouched.

Training provenance

Base model: Qwen/Qwen2.5-0.5B-Instruct (Apache-2.0)
Adapter: Thox-ai/ThoxAir-0.5B-WaveC2 (fp LoRA r32/alpha 64 on the 7-module projection list, base loaded fp16, NOT quantized; train_edge_sft.py, Thox-ai/thox-edge-sft @ c250f9b, adapted for Wave C.2; LR 1e-4 cosine, warmup 0.05, 3 epochs, per-device batch 8, max_length 1024)
Dataset: Thox-ai/thox-factory-sft-v5 @ train_wave_c2.jsonl (1,246 rows, commit 7953c54355ea45e9f30d8d40c71ecdf5911f2396, SHA-256 771434812b6a14fceba7f929fcbe1040d0e6be7a32c9495c1e8d0784512919a3)
Training job: 6ae40f040e0338d1bb7c7944 (HF Jobs, a10g-small; 468 steps, final train loss 0.6478)
Gate job: 6ae414050e0338d1bb7c7c35 (full per-question report in the adapter repo's thox_gate49.json)
Merge: bf16 merge_and_unload on the finishing VM (transformers 5.21.0, peft 0.21.2, torch CPU). Merged weights commit 595b779a80a3e90d0f52a0f7953265d70e844349 in the source repo; model.safetensors 988,097,824 B, sha256 2497a347a4ac0a9b92e6e6be50ed6fb5b2c7d1bb2c174b9c1f1aa04f3e1433fb.

GGUF and Ollama artifacts

Converted from these merged weights (llama.cpp convert_hf_to_gguf.py + llama-quantize; the f16 intermediate is retained on the build host):

ThoxAir-0.5B.Q4_K_M.gguf — 397,807,584 B, sha256 fb32e5ab802df466809836182f071bb0ab0871abb9273131456c89d998bccd6a
ThoxAir-0.5B.Q8_0.gguf — 531,067,872 B, sha256 936a144c99ec2f61d45832171e0079e3275facec4c0e89bb6ed5ccc8a771b328
Modelfile builds the Ollama artifact from the Q4_K_M GGUF: ChatML template, the grounding system prompt verbatim, temperature default 0 — the gated configuration.

Limitations

A 0.5B model: narrow capability by design; the adapter is a facts and behaviour skin on the base, not a capability upgrade. Expect correct declines on hard questions rather than correct answers.
Small models are the most fabrication-prone tier in the fleet's gate history; the gate result on this card is the claim, nothing more.
Pair deployed surfaces with MeshStack escalation for questions outside the trained facts.

License

Apache-2.0, inherited from the base model (Qwen/Qwen2.5-0.5B-Instruct, Apache-2.0). The adapter is a derivative and cannot be freer than its base.

Back THOX on Kickstarter

THOX.ai is live on Kickstarter. Explore ThoxKey, ThoxAir, ThoxMini and ThoxClip, and help bring the THOX personal AI ecosystem to life.

Back THOX on Kickstarter

Citation

bibtex

@misc{thoxai2026thoxair05b,      title={ThoxAir-0.5B},      author={THOX.ai},      year={2026},      url={https://huggingface.co/Thox-ai/ThoxAir-0.5B}}

@misc{thoxai2026thoxair05b,      title={ThoxAir-0.5B},      author={THOX.ai},      year={2026},      url={https://huggingface.co/Thox-ai/ThoxAir-0.5B}}