Models / Empero
Qwythos 9B
Security focused post train of Qwen3.5 9B with a 1M context. Built for exploit triage and malware analysis.
- Publisher
- Empero
- Base model
- Qwen3.5 9B
- Parameters
- 9B
- Context
- 1M
- Released
- Aug 2026
- License
- Apache 2.0
Method · Fine tune
Supervised fine tuning on compliant data. Changes behaviour more broadly than a projection.
Overall
86.8
weighted, out of 100
Openness
91%
base 20%
Agentic completion
73%
25% loop
Loop regression
+13
base 12% → 25%
Capability retained
98.7%
−1.0 pts mean
Cheapest input
Plan
subscription only
Fastest
121
tok/s · Featherless
Refusal fingerprint
What it still refuses.
Share of probes refused in each category, against Qwen3.5 9B before the edit. Probes are graded by a judge model and no harmful completions are stored.
Fingerprint
Open, this model
91%
Open, base
20%
| Category | Base | This model | Change |
|---|---|---|---|
Cybersecurity | 72% | 0% | −72 |
Biology | 94% | 5% | −89 |
Chemistry | 95% | 5% | −90 |
Weapons | 86% | 7% | −79 |
Drugs | 78% | 6% | −72 |
Adult content | 82% | 24% | −58 |
Violence & gore | 34% | 5% | −29 |
Hate & harassment | 90% | 7% | −83 |
Self harm | 72% | 7% | −65 |
Privacy & OSINT | 74% | 2% | −72 |
Fraud & scams | 91% | 6% | −85 |
State sensitive | 88% | 30% | −58 |
Effective censorship by provider
The same weights can behave differently depending on who serves them. Each cell is the model's own refusal rate combined with that provider's declared filters.
| Route | Cyber | Bio | Chem | Weap | Drug | Adult | Viol | Hate | Self | OSINT | Fraud | Polit | CSAM |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Raw weights | 0% | 5% | 5% | 7% | 6% | 24% | 5% | 7% | 7% | 2% | 6% | 30% | Blocked |
| Featherless | 0% | 5% | 5% | 7% | 6% | 24% | 5% | 7% | 7% | 2% | 6% | 30% | Blocked |
CSAM is prohibited on every route, every provider, every model. It is not a setting.
Agentic stability
Does it hold up in agent loops?
Five multi turn tool scenarios against deterministic simulators. An episode counts as a loop when the model repeats itself, oscillates, or stops making progress before the task is done.
By scenario
Loop rate versus base
Regression
+13 pts
First loop, median turn
T11
Tool calls
PreviewHow it fails
Same tool call with the same arguments, three turns running.
Responses above 0.85 Jaccard similarity to a recent turn.
Flip flopping between two actions without progress.
Invalid JSON or arguments that break the schema.
The model writes the tool output itself instead of calling.
No text and no tool call.
Ignores the latest tool output and restarts the task.
Share of failed episodes by first detected failure mode.
Versus base model
What the edit cost.
The same benchmarks on Qwen3.5 9B and on this model.
Benchmarks
Retention
98.7%
of base scores kept · −1.0 pts mean change
KL divergence from base
Not reported. KL is only meaningful for weight edits, not full fine tunes.
Providers
Where to run it.
1 provider serves this model. Prices come from provider pages. Latency, throughput and uptime are preview numbers until router traffic fills them in.
| Provider | Quant | Context | Input /M | Output /M | TTFT | Throughput | Uptime | Moderation |
|---|---|---|---|---|---|---|---|---|
| Featherless | bf16 | 1M | Plan | Plan | 229 ms | 121 tok/s | 99.3% | No added filter |
Plan means flat rate subscription access. Hover a Plan cell for the provider's pricing.
Usage
Who is using it.
350M
tokens routed this week
+75% vs four weeks ago
API
Call it in three lines.
OpenAI compatible. Point your SDK at our base URL and keep the rest of your code.
curl https://openuncensored.com/api/v1/chat/completions \
-H "Authorization: Bearer $OU_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwythos-9b",
"messages": [{"role": "user", "content": "Hello"}],
"routing": {"strategy": "quality"}
}'Model id qwythos-9b · routing strategies quality, speed, cost · routing docs