Archive · vps-81 historical telemetry · report-catalog.html. Originally rendered 2026-04-18. Re-hosted from MyServers on 2026-05-06. Methodology and harness conventions may differ from what we use today; see /methodology.html for current standards. ← back to all benchmarks
Telemetry Overview

Telemetry Report Catalog

This page is now model-first. The sidebar lists every model we have local evidence for, each model page collects its findings and linked benchmark pages, and the main body highlights the most relevant reports across the Mac lane, the Windows lane, the VPS lane, and the concurrent benchmark family.

Spotlight

Recent and Important Reports

Python 10Q

vps50 Python Telemetry Mini Report

Qwen3.5 4B

Python 20Q

vps50 Python Telemetry Mini Report

Qwen3.5 4B

Small Eval

vps50 Small-Model Telemetry Manual

Qwen3.5 4B

Hello Check

Hello Check Report

Qwen3.5 4B

Windows GPU

2026-04-13 Pavilion Windows GPU Qwen3.5 0.8B Long Context

Qwen3.5 0.8B

Report

Qwen3.5 9B Local Mac Progress

Qwen3.5 9B

Python 20Q

vps50 Python Telemetry Mini Report

Qwen3.5 9B

Small Eval

vps50 Small-Model Telemetry Manual

Qwen3.5 9B

Windows GPU

2026-04-13 Pavilion GPU lane - Qwen3.5 2B

General telemetry page

Windows GPU

2026-04-13 Pavilion Windows GPU Qwen3.5 Long Context

General telemetry page

Models

Model Directory

Use this grid when you want the short version first. Each card links to a model-dedicated page with findings, model details, and benchmark links.

model

CodeLlama 34 16k

Latest 10Q suite on ollama averaged 415.3s for primary answers with 10/10 usable primary responses.

model

Codestral 32k

Latest 10Q suite on ollama averaged 292.5s for primary answers with 10/10 usable primary responses.

model

Llama 3.2 3B

Latest small-eval run averaged 62.7s per question, 2/5 strict passes, and 5.76 tok/s throughput.

model

Phi-3 Mini

Latest small-eval run averaged 52.4s per question, 4/5 strict passes, and 6.36 tok/s throughput.

model

Phind 34 16k

Latest 10Q suite on ollama averaged 405.9s for primary answers with 10/10 usable primary responses.

model

Qwen14 Coder 32k

Latest 10Q suite on ollama averaged 231.0s for primary answers with 10/10 usable primary responses.

model

Qwen14 General 32k

Latest 10Q suite on ollama averaged 156.6s for primary answers with 10/10 usable primary responses.

model

Qwen2.5 3B

Latest small-eval run averaged 59.1s per question, 3/5 strict passes, and 5.57 tok/s throughput.

qwen2.5-coder

Qwen2.5 Coder 0.5B

Latest hello-check finished in 52.9s on the recorded runner.

qwen2.5-coder

Qwen2.5 Coder 1.5B

Latest hello-check finished in 4.8s on ollama-local-mac.

qwen2.5-coder

Qwen2.5 Coder 14B

Latest hello-check finished in 138.3s on ollama-local-mac.

qwen2.5-coder

Qwen2.5 Coder 3B

Latest hello-check finished in 5.8s on ollama-local-mac.

qwen3.5

Qwen3.5 0.8B

Use the linked model page to browse the available reports.

qwen3.5

Qwen3.5 4B

Latest hello-check finished in 14.4s on ollama-local-mac.

qwen3.5

Qwen3.5 9B

Latest hello-check finished in 4643.4s on ollama-local-mac.

model

Qwen32 Coder 32k

Latest 10Q suite on ollama averaged 418.9s for primary answers with 10/10 usable primary responses.

smollm2

SmolLM2 1.7B

Use the linked model page to browse the available reports.

smollm2

SmolLM2 135M

Use the linked model page to browse the available reports.

smollm2

SmolLM2 360M

Use the linked model page to browse the available reports.

smolvlm

SmolVLM 500M

Use the linked model page to browse the available reports.

smolvlm2

SmolVLM2 256M Video

Use the linked model page to browse the available reports.