local

/home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf

Released Context 0Khome-thomaskyn-models-qwen3.5-4b-qwen3.5-4b-q4_k_m.gguf

PipelineScore

57.8TAP

Ranked #12 of 13 models · 8th percentileTool use is the headline (87.5); throughput is the soft spot (29.2). Best-fit profile: Agentic.

Category breakdown

Score per category, normalized 0–100 against the v1 anchor.

Code

50.0

Reason

40.0

Tool Use

87.5

RAG

87.5

Speed

29.2

Strengths

Tool Use87.5

RAG87.5

Code50.0

Same model, different rigs

Every submission of /home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf on the 0–100 scale. The spread is the point: where it runs changes what you get.

0255075100

Best 57.8 on gtx-1060-5gb-5gb · lowest 57.8 on gtx-1060-5gb-5gb · spread 0.0 pts across 2 runs. Hover a dot for its rig.

Sample tasks

A taste of what the test pack measures. Full prompts are private and rotated daily.

CodeDifficulty 1code-fib-1

Fibonacci function

Write a Python `fib(n)` returning the nth Fibonacci number, O(n).

ReasonDifficulty 1reason-math-1

Train meeting time

Two trains, opposite directions, given speeds and start times — when do they meet?

RAGDifficulty 2rag-extract-1

Extract metrics to JSON

From the context, extract net sales, operating margin, and free cash flow as a JSON object. Numbers only.

Tool UseDifficulty 2tool-schema-1

OpenAPI param selection

Given an OpenAPI schema with limit/offset/sort, fill JSON for 'next 50, recent first.'

RAGDifficulty 2rag-grounding-1

Refuses to fabricate

Context lacks the answer — does the model fabricate or correctly say it can't?

Compare with

/home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf vs openai/gpt-oss-20b /home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf vs gpt-oss:latest /home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf vs qwen3.6-35b-a3b /home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf vs qwen3-coder:30b /home/thomaskyn/models/Qwen3.5-4b/Qwen3.5-4B-Q4_K_M.gguf vs qwen3.8:27b