Will It Run AI

MistralMistral

Devstral 2 123B Instruct

Frontier
huggingfaceHuggingFace
10.2KDownloads322LikesNov 2025Veröffentlicht256K TokenKontextMistral Research LicenseLizenz100 HerausragendQualität

Devstral 2 123B Instruct (123B parameters) requires approximately 81.9 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 95 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Devstral 2 123B Instruct on your machine.

Run

lms load Devstral-2-123B-Instruct-2512 && lms server start

Quick specs

Parameters123B
Architecturedense
Context256K tokens
Modalitytext+vision
Min RAM48 GB
Rec. RAM75 GB (Q4_K_M)
LicenseMistral Research License
FamilyDevstral
✓ Vision✓ Code✓ Reasoning

About this model

Devstral is an agentic LLM for software engineering tasks. Devstral 2 excels at using tools to explore codebases, editing multiple files and power software engineering agents. The model achieves remarkable performance on SWE-bench.

  • •Agentic Coding: Devstral is designed to excel at agentic coding tasks, making it a great choice for software engineering agents
  • •Improved Performance: Devstral 2 is a step-up compared to its predecessors
  • •Better Generalization: Generalises better to diverse prompts and coding environments
  • •Context Window: A 256k context window

Deine Hardware

Erkennung...

Schnellauswahl

Bestes BudgetS
Mac Studio M3 Ultra 256GB~$6,999 — 8 tok/s
Beste GesamtwahlS
AMD Instinct MI300A 128GB~$12,000 — 54 tok/s

Beste Hardware

Top-Empfehlungen für Devstral 2 123B Instruct

AMD Instinct MI300A 128GBS
128 GB
NVIDIA
NVIDIA H200 141GBS
141 GB
NVIDIA
NVIDIA H200 PCIe 141GBS
141 GB
Intel
Gaudi 3 128GBS
128 GB
AMD Instinct MI250X 128GBS
128 GB

Dieses Modell ausführen

Devstral 2 123B Instruct on AMD Instinct MI300A 128GBDevstral 2 123B Instruct on NVIDIA H200 141GBDevstral 2 123B Instruct on NVIDIA H200 PCIe 141GB

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

QuantBitsVRAMQualityFit
Q2_K
2
48.0 GB
Low—
Q3_K_S
3
60.3 GB
Low—
NVFP4
4
68.9 GB
Medium—
Q4_K_M
4
75.0 GB
Medium—
Q5_K_M
5
88.6 GB
High—
Q6_K
6
100.9 GB
High—
Q8_0
8
131.6 GB
Very High—
F16
16
252.2 GB
Maximum—

Quality benchmarks

Devstral 2 123B Instruct benchmark scores

Benchmark verified

Coding

SWE-bench Verified72.2%
HumanEval+—
Aider Polyglot—
LiveCodeBench—

Source: official · 2025-12-18

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights75.0 GB
KV Cache5.4 GB
Runtime0.9 GB
Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Devstral 2 123B Instruct

Siehe auch

QuantisierungsleitfadenBewertungsmethodikRechner öffnen