Devstral-Small-2-24B-Instruct-2512 on AMD GPUs

17 results · 2 sources · 54 measurements

Grouped by reported model identity. Exact weights, quantizations and software can differ. How results are reviewed

Documented results

17 selected tests


View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	HumanEval100 problems	88 %Reported pass@1 · n = 100	—	31.57 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / ? tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportLocalMaxxing	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗	Token speed	—	566.77 tok/sPrefill · median · unknown	37.67 tok/sGeneration · median · unknown	1 × R9700Single GPU · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS linux Engine llama.cpp Backend rocm Harness LocalMaxxing community speed test; exact client where reported Prompt / output 41 / 82 tokens Context 16384 Concurrency Not reported Date 2026-09-21 Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-bulk-dispatcher-impl	—	—	31.35 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 836 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportLocalMaxxing	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗	Token speed	—	580.12 tok/sPrefill · median · unknown	37.87 tok/sGeneration · median · unknown	1 × R9700Single GPU · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS linux Engine llama.cpp Backend rocm Harness LocalMaxxing community speed test; exact client where reported Prompt / output 41 / 87 tokens Context 16384 Concurrency Not reported Date 2026-09-28 Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-horner-mpmath-derivative	—	—	31.56 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 266 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportLocalMaxxing	Devstral-Small-2-24B-Instruct-2512Q8_0Model ref. ↗	Token speed	—	—	23.87 tok/sGeneration · peak · unknown	1 × R9700Single GPU · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS Arch Linux Engine llama.cpp Backend vulkan Harness LocalMaxxing community speed test; exact client where reported Prompt / output 1078 / 139 tokens Context 24576 Concurrency Not reported Date 2026-05-05 Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-invoke-sign-artifacts-ps	—	—	31.39 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 670 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportLocalMaxxing	Devstral-Small-2-24B-Instruct-2512Q8_0Model ref. ↗	Token speed	—	—	23.56 tok/sGeneration · peak · unknown	1 × R9700Single GPU · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS Arch Linux Engine llama.cpp Backend vulkan Harness LocalMaxxing community speed test; exact client where reported Prompt / output 1078 / 135 tokens Context 1213 Concurrency Not reported Date 2026-05-06 Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-legacy-rename-toolchain	—	—	31.9 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 553 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-nuget-multitarget-design	—	—	31.6 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 308 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-parser-nunit-tests	—	—	31.4 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 580 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-poincare-section-pytest	—	—	31.33 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 709 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-range-expand-bug-fix	—	—	31.7 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 109 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-relay-command-refactor	—	—	31.49 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 292 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-resx-toolbar-button	—	—	31.52 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 332 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-vanderpol-rk4-bugfix	—	—	31.66 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 139 tokens Context Not reported Concurrency Not reported Date Not reported Full record →
View ReportCustom evaluation	Devstral-Small-2-24B-Instruct-2512Q4_K_MModel ref. ↗Conflicting evidence	Token speedsynth-vanderpol-rk4-stepper-c	—	—	31.46 tok/sGeneration · mean · per request	2 × W7800Layer split · 48 GB VRAM/GPU Configuration Memory 48 GB VRAM/GPU OS Ubuntu 24.04 LTS Engine Ollama Backend ROCm Harness BoJl unified evaluation harness (2026-08) Prompt / output ? / 507 tokens Context Not reported Concurrency Not reported Date Not reported Full record →

Quantizations and software

Quantizations: Q4_K_M, Q8_0
Engines: llama.cpp, Ollama

Explore all 17 results →

This is a selection across sources, not a ranking or a matched experiment. Reading benchmark reports

GPU benchmark pages: W7800 · R9700