The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Underdog Saluki 27B 1.0 does score higher than full-size Qwen3.8-27B on its publisher’s tool-calling test: 88 of 120 tasks versus 84. That is a modest result from a test run by the model’s publisher, not proof that Saluki is generally better. Its 7.89 GB IQ2-mix GGUF is designed for local inference with llama.cpp, but the model card reports weaker results on several math and reasoning benchmarks and warns of formatting slips in parallel tool calls.
What Underdog Saluki 27B is
Underdog Saluki 27B 1.0 is a compact GGUF quantization based on Qwen3.8-27B. ConwayResearch lists the main file, Underdog-Saluki-27B-1.0-IQ2-mix.gguf, at 7.89 GB and licenses it under Apache 2.0. Its model card describes the release as “Qwen3.8-27B in under 8 GB, tuned to keep tool calling intact.” (ConwayResearch model card)
As an Amazon Associate I earn from qualifying purchases.
“2-bit” refers to the IQ2-mix quantization designation, not a claim that every part of the file uses a uniform two bits per parameter. The smaller file is intended to make local use more practical than the full-size model, while quantization can involve quality trade-offs.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Does Saluki really beat the original at tool calling?
In ConwayResearch’s Underdog Bench, Saluki passed 88 of 120 tasks derived from BFCL v4; full-size Qwen3.8-27B passed 84, and Bonsai 2 passed 70. The publisher says tasks were frozen before testing, thinking was disabled, and temperature was set to 0. The four-task gap favors Saluki in this particular evaluation, but the card calls the test modest and notes that a few tasks’ difference could reflect run-to-run variation. The consulted sources do not independently reproduce the result. (ConwayResearch model card)
#1 Best Overall
- GPU-Modell: Gefoce RTX 3080
- Memory Type: GDDR6X Memory Capacity: 20GB Memory Bus Width: 320bit Output Interfaces: 3*DP + HDMI Core Clock: 1710MHz Memory Clock: 19Gbps Power Interface: 8+8pin Recommended Power Supply: 850W or higher
That distinction matters: this is evidence of a promising result on one creator-run test set, not evidence that Saluki will outperform the base model across tools, prompts, runtimes, or real workloads.
How the other reported results compare
The model card reports several additional scores. Some favor Saluki, while math and reasoning results favor the full-size model. The public comparison figures below were run with a different harness, according to the card, so they should not be read as controlled, directly comparable head-to-head measurements.
Rank #2
- [Wireless Mobile Mini Travel Router] The NanoPi M5 mini router is an open-sourced mini smart gateway device, designed and developed by FriendlyElec. It is based on Rockchip RK3576 SoC, with 32-bits LPDDR4X/LPDDR5 RAM and UFS 2.0 storage(optional). The RK3576 is an 8-core 64-bit processor featuring a powerful architecture with 4x ARM Cortex-A72 cores and 4x ARM Cortex-A53 cores. It is equipped with an ARM Mali G52 MC3 GPU and 6 TOPS NPU.
- [Greater Storage and Scalability]] NanoPi M5 Portable Wireless Mini Router onboard 4GB LPDDR4X/ 8GB 16GB LPDDR5 RAM. On-Board 16MB SPI Nor flash Supports microSD up to UHS-I Supports UFS 2.0 flash module. Supports M.2 M-Key 2280 NVMe SSD (PCIe 2.1 x1). 2x one Gbps Ethernet ports with RTL8211F PHY chips Supports M.2 SDIO Wi-Fi/BT module. 2x USB 3.2 Gen 1 Type-A ports. 30-Pin 2.54mm GPIO header. 2x 4-Lane MIPI CSI-2 D-PHY v1.2 interfaces.
- [Al Model Performance] Nanopi M5 Mini Router support Al Model Performance and Resource Usage on. Supporting Local Deployment & Running of Al Models, such as Llama-3.2, Chat GLM3, Deep Seek R1, Int ern LM2, Qwen 2.5 and so on mainstream AI inference modeling platforms.It is very suitable for enterprise customers to customize the development of mini machine vision systems with multiple network ports.
- [Open Source and Programmable] NanoPi M5 computer mini wifi router can support FriendlyWrt OS, a custom system based on the OpenWrt distribution. It is open source and ideal for developing IoT applications, NAS applications, smart home office gateways and more. NanoPi M5 mini wifi router can support external USB wifi adapter. Simultaneous dual band and Convert a public network(wired/wireless) to a private Wi-Fi for secure surfing.
- [Wide Range of Operating Systems] NanoPi M5 Portable Wireless Mini Router running Android 14 Tablet, Android 14 TV, Debian 11 Desktop, FriendlyWrt 21.02, FriendlyWrt 23.05, FriendlyWrt 24.10, OpenMediaVault OS System. Also support Proxmox VE, Ubuntu 20.04 Desktop, Ubuntu 24.04 Core and Ubuntu 24.04 Desktop. Kernel version: Linux-6.1-LTS and U-boot-2017.09.It is also fully compatible with headless systems.
| Evaluation | Saluki | Full-size Qwen3.8-27B | What the card says |
|---|---|---|---|
| Parallel BFCL v4 tasks (100) | 42 | 35 | Publisher test; official checker, thinking disabled |
| SWE-bench Verified (50 issues) | 30 | 33 | Card-reported results |
| IFEval, prompt-loose | 93.5 | 91.5 public | Public comparison uses a different harness |
| IFBench, prompt-loose | 72.7 | 71.0 public | Public comparison uses a different harness |
| MBPP+ | 78.0 | 83.9 public | Public comparison uses a different harness |
| MuSR | 67.5 | 79.6 public | Public comparison uses a different harness |
| AIME 2025, avg@4 | 79.2 | 96.7 public | Public comparison uses a different harness |
| AIME 2026, avg@4 | 80.0 | 94.6 public | Public comparison uses a different harness |
The parallel-call test is also publisher-run and covers 100 BFCL v4 tasks. Although Saluki scores 42 to the full model’s 35, ConwayResearch cautions that about one fifth of Saluki’s parallel-call replies have small formatting slips. A higher task score therefore does not mean every multi-call response will be cleanly formatted for downstream software.
The card characterizes Saluki’s competition-math performance as about 82–85% of the full model’s, identifies letter-level instruction puzzles as a weak spot, and says that enabling thinking can lead to long reasoning before an answer. These are useful cautions if the intended workload depends on exact constraint-following or strong math reasoning.
Rank #3
- [Wireless Mobile Mini Travel Router] The NanoPi M5 mini router is an open-sourced mini smart gateway device, designed and developed by FriendlyElec. It is based on Rockchip RK3576 SoC, with 32-bits LPDDR4X/LPDDR5 RAM and UFS 2.0 storage(optional). The RK3576 is an 8-core 64-bit processor featuring a powerful architecture with 4x ARM Cortex-A72 cores and 4x ARM Cortex-A53 cores. It is equipped with an ARM Mali G52 MC3 GPU and 6 TOPS NPU.
- [Greater Storage and Scalability]] NanoPi M5 Portable Wireless Mini Router onboard 4GB LPDDR4X/ 8GB 16GB LPDDR5 RAM. On-Board 16MB SPI Nor flash Supports microSD up to UHS-I Supports UFS 2.0 flash module. Supports M.2 M-Key 2280 NVMe SSD (PCIe 2.1 x1). 2x one Gbps Ethernet ports with RTL8211F PHY chips Supports M.2 SDIO Wi-Fi/BT module. 2x USB 3.2 Gen 1 Type-A ports. 30-Pin 2.54mm GPIO header. 2x 4-Lane MIPI CSI-2 D-PHY v1.2 interfaces.
- [Al Model Performance] Nanopi M5 Mini Router support Al Model Performance and Resource Usage on. Supporting Local Deployment & Running of Al Models, such as Llama-3.2, Chat GLM3, Deep Seek R1, Int ern LM2, Qwen 2.5 and so on mainstream AI inference modeling platforms.It is very suitable for enterprise customers to customize the development of mini machine vision systems with multiple network ports.
- [Open Source and Programmable] NanoPi M5 computer mini wifi router can support FriendlyWrt OS, a custom system based on the OpenWrt distribution. It is open source and ideal for developing IoT applications, NAS applications, smart home office gateways and more. NanoPi M5 mini wifi router can support external USB wifi adapter. Simultaneous dual band and Convert a public network(wired/wireless) to a private Wi-Fi for secure surfing.
- [Wide Range of Operating Systems] NanoPi M5 Portable Wireless Mini Router running Android 14 Tablet, Android 14 TV, Debian 11 Desktop, FriendlyWrt 21.02, FriendlyWrt 23.05, FriendlyWrt 24.10, OpenMediaVault OS System. Also support Proxmox VE, Ubuntu 20.04 Desktop, Ubuntu 24.04 Core and Ubuntu 24.04 Desktop. Kernel version: Linux-6.1-LTS and U-boot-2017.09.It is also fully compatible with headless systems.
Running it locally with llama.cpp
The model card documents stock llama.cpp and shows a server setup using --jinja, GPU-layer offload, flash attention, and a 32,768-token context. It says --jinja enables the Qwen3.8 chat template used for tool calls and thinking. These are example settings from the card, not a guaranteed hardware requirement or a performance promise; the source does not specify a minimum computer configuration. (ConwayResearch model card)
Check the model card for its current command and setup details before running the server. In particular, confirm that the llama.cpp build and chosen options support the model’s chat template and the tool-calling flow you intend to use.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Vision is a separate add-on
The main GGUF is text-only. For vision, the card lists an optional F16 add-on of 928 MB or a Q8_0 add-on of 629 MB, passed with --mmproj in its documented setup. Those files are additional to the main model; having Saluki’s text model alone does not provide vision input.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsWho should consider Saluki?
Saluki makes most sense for someone who wants to try a smaller local Qwen3.8-27B-derived model and values tool calling enough to test it in their own workflow. Compare the trade-offs before choosing:
- Consider Saluki if the smaller 7.89 GB model file is useful and your workload is tool-centric; verify tool-call formatting with the actual application that will consume the responses.
- Prefer the full-size model or test both if competition math, reasoning, or reliably exact instruction puzzles are central to your use case.
- Do not treat the benchmark lead as a guarantee if your prompts, tools, sampling settings, or runtime differ from the publisher’s evaluation.
The benchmark numbers and setup details here are claims and specifications from the ConwayResearch Hugging Face model card. The secondary MarkTechPost listing is dated October 9, 2026, but the benchmark evidence discussed above comes from the model card, not an independent test.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
