🔍 Read the full analysis: Could A Small Language Model Handle Your Personal Assistant Tasks? on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Tether AI Research has introduced BotPsy, a 0.8-billion-parameter model based on Qwen3.5-0.8B and designed for tool use and assistant workflows, including on-device deployment. The company reports gains on six benchmarks, but those results have not been independently verified and do not establish how reliably BotPsy will handle everyday tasks.
Tether AI Research has introduced BotPsy, a 0.8-billion-parameter model built on Qwen3.5-0.8B for tool use and agent workflows, including potential on-device deployment, as detailed in the original analysis. The company reports improved results over the base model across several benchmarks, but no independent evaluation or user deployment data is provided, so the scores do not show whether BotPsy can reliably manage personal assistant tasks in everyday use.
BotPsy is designed to call tools for tasks including web research, travel requests, commerce, calendars, messaging, document handling and device controls. Tether AI Research describes it as suited to low- and medium-difficulty work, rather than high-stakes decisions or frontier-level general-purpose tasks. The company also cautions that the model can make mistakes.
In results reported by the developer, BotPsy scored 44.0 on BFCL v4, compared with 23.4 for the Qwen3.5-0.8B-Think base model. It scored 58.7 versus 30.1 on ACEBench, and 51.5 versus 23.2 on τ-bench. Tether says the τ-bench score is the mean of its retail, telecom and banking subsets. The source says BotPsy led models with up to 1.2 billion parameters on six tool-use benchmarks, though it does not provide independent confirmation of that comparison.
Results against the larger Qwen3.5-2B-Think were mixed. BotPsy scored higher on five of the six listed tests, including BFCL v4, ACEBench, τ-bench, ToolSandbox and BrowseComp-Plus. The larger model scored 48.9 on Claw-Eval, compared with 46.5 for BotPsy. The announcement also reports gains in instruction-following and some knowledge tests, while flagging weaker performance as instructions accumulate and in lower-resource languages.
BotPsy brings tool use to a compact model
Tether AI Research says its small assistant is designed for routine agent workflows and on-device use. Its benchmark results are developer-reported and do not establish everyday reliability.
Built for routine assistant work
The model is tuned for function calling and agent workflows, rather than frontier-scale general performance.
Find and gather information
Example uses include web research and travel requests.
Handle everyday workflows
It is presented for commerce, business operations, documents, calendars, and messaging.
Run on-device
Local deployment could reduce reliance on cloud services for routine tasks.
Reported results improved over its base
Tether AI Research reports higher scores for BotPsy on three named benchmarks.
| Benchmark | BotPsy | Qwen3.5-0.8B-Think |
|---|---|---|
| BFCL v4 | 44.0 | 23.4 |
| ACEBench | 58.7 | 30.1 |
| τ-bench | 51.5 | 23.2 |
The larger-model comparison was mixed
BotPsy scored higher on five of six tool-use tests against Qwen3.5-2B-Think.
- BFCL v4
- ACEBench
- τ-bench
- ToolSandbox
- BrowseComp-Plus
- Qwen3.5-2B-Think scored 48.9 on Claw-Eval, compared with BotPsy’s 46.5.
Limits matter for deployment
The supplied material describes areas where performance may fall short and leaves practical questions open.
- Results come from the developer; independent reproduction is not provided.
- Performance drops as instructions accumulate and lags in lower-resource languages.
- Hardware needs, memory use, power use, and real-world reliability are not established.
- Tether says BotPsy can make mistakes and is not intended for high-stakes use.
Local Assistants Face a Reliability Test
A compact model that can select and use tools could let some assistant tasks run directly on a phone or computer, rather than sending every request to a cloud service. If BotPsy can perform reliably on suitable devices, that could give users more control over where some requests and data are processed. The announcement, however, does not document actual deployments or establish that those privacy benefits occur in practice.
The reported safety-related figures also need careful interpretation. Tether says BotPsy scored 86.1 on BFCL v4 hallucination tests assessing whether it abstains when no suitable tool is available. It also reports a 31% unsafe-response rate under prompt-injection attacks, down from 46% for the base model. These are developer-reported test results, not proof of safety across real-world tools, users or attack conditions.
For people considering a personal assistant, the practical test is not benchmark ranking alone. It is whether the model can follow requests accurately, recover when a tool fails, protect sensitive information and avoid taking unintended actions. The available material does not answer those questions.
Products Worth Considering
Amazon Kindle Paperwhite (2024, 16 GB) – 7-inch Glare-Free Display, Black
LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz
DJI Mini 3 Drone with DJI RC and 3-Axis Mechanical Gimbal
As an affiliate, we earn on qualifying purchases.
A Tool-Use Model Built on Qwen
BotPsy is based on Qwen3.5-0.8B and is distributed as model weights, with the source listing safetensors and GGUF formats in bf16, Q8 and Q4. Its stated focus is function calling and agent workflows: the model is meant to choose and invoke tools, rather than serve as a general-purpose model expected to match much larger systems across all tasks.
Tether’s rationale is that smaller, specialized models may handle routine work with lower operating costs and more local data control than cloud-based alternatives. That is the developer’s case for the approach, not a verified cost comparison. The supplied material contains selected benchmark figures but no independent tests, consumer deployment reports or evidence of performance across different devices.
“BotPsy is a “personal assistant” that uses tools to perform a variety of simple tasks for the user.”
— Tether AI Research
Benchmarks Leave Practical Questions
The reported scores come from Tether AI Research; the supplied material does not identify an independent evaluator or give full testing protocols. It is unclear whether the results will reproduce across different runtimes and implementations, or how well benchmark performance will transfer to a personal assistant handling changing tools and varied requests.
The source does not specify BotPsy’s memory use, power consumption, latency or performance on particular devices. It also gives no end-user cost comparison with cloud models. Although it says results weaken as instructions accumulate and in lower-resource languages, the provided details do not quantify those limitations. The announcement does not state a release date or describe support and update commitments.
Likewise, the reported prompt-injection and abstention results do not establish how the model behaves with sensitive information, real services or longer sequences of actions. The extent of safeguards available to people deploying it remains unclear from the source material.
Independent and Device Tests Needed
The next meaningful evidence would be independent reproduction of the benchmark results and tests on representative phones, computers and other devices. Such evaluations could report memory requirements, latency, energy use and reliability when tools fail or a workflow runs across multiple steps.
People considering deployment will also need documentation on supported runtimes, safeguards, updates and the limits of the model’s language and instruction-following performance. Tether AI Research has not provided a timeline for further evaluations or releases in the supplied material, so no specific next milestone is confirmed.
Key Questions
What is BotPsy?
BotPsy is a 0.8-billion-parameter model based on Qwen3.5-0.8B. Tether AI Research designed it for tool use and assistant workflows, including tasks such as research, calendar handling and messaging.
Can BotPsy run on a personal device?
Tether presents BotPsy as suitable for on-device deployment and lists model-weight formats including GGUF. The supplied information does not report performance, memory use or power needs on specific devices.
Are BotPsy’s benchmark scores independently verified?
No independent evaluation is identified in the supplied material. The figures are reported by Tether AI Research, and the source does not provide full testing protocols.
Is BotPsy intended for high-stakes tasks?
No. Tether AI Research says the model is aimed at low- to medium-difficulty tasks, is not intended for high-stakes use and can make mistakes.
When will BotPsy be available, and what will it cost?
The supplied source gives no release date or end-user pricing, and does not compare operating costs with cloud-based assistants.
Primary source: Hugging Face · via ThorstenMeyerAI.com
