Back to leaderboard

mistralai/devstral-small-2-2512

LM-STUDIO MLX

mistral3 · 24B · 4bit

Good

Oct 20, 2026 · Apple M1 Pro

granite-3.3-2b-instruct

LM-STUDIO MLX

granite · 2B · bf16

Good

Mar 4, 2026 · Apple M4

Global Score
73 vs 66
Hardware Fit
64 vs 87
Quality Score
77 vs 57

Hardware

mistralai/devstral… granite-3.3-2b-ins…
MachineMacBook ProMacBook Air
CPUApple M1 ProApple M4
Cores1010
RAM32 GB32 GB
GPUApple M1 ProApple M4
OSmacOS 15.6.1macOS 26.3
Archarm64arm64
Power Modebalancedbalanced

Performance

mistralai/devstral… granite-3.3-2b-ins…
Tokens/sec11.918.3
First chunk7 ms541 ms
TTFT1.1 s541 ms
Load timeN/AN/A
Memory usage19.7 GB4.7 GB
Memory %62%15%

HW Fit Score Breakdown

mistralai/devstral-smal…

Speed
25/50
TTFT
20/20
Memory
19/30

granite-3.3-2b-instruct

Speed
37/50
TTFT
20/20
Memory
30/30

Quality

mistralai/devstral-smal…

Reasoning
17/20
Coding
16/20
Instruction
13/20
Structured
14/15
Math
7/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Adequate Structured Output: Strong Math: Weak Multilingual: Strong

granite-3.3-2b-instruct

Reasoning
10/20
Coding
10/20
Instruction
14/20
Structured
14/15
Math
2/15
Multilingual
7/10
Reasoning: Adequate Coding: Weak Instruction Following: Adequate Structured Output: Strong Math: Poor Multilingual: Adequate

Run yours and compare

$ npm install -g metrillm@latest
$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest