Model comparison

Codestral vs Llama 3.2 3B

Codestral is the stronger model overall, scoring 30.6 to 28.9 on the Noometry Index. Llama 3.2 3B costs 3.7× less per token, which makes it the better buy when Codestral's lead doesn't matter for your workload.

Last verified October 10, 2026. 2 shared benchmarks.

Codestral

Mistral AI

30.6

Rank #290 Reported

Llama 3.2 3B

Summary

They share 2 benchmarks with published results for both. Codestral scores higher in 0 categories and Llama 3.2 3B in 2 categories; one gap is clear of the uncertainty.
The biggest single-benchmark swing is BigCodeBench Complete: 52.5% for Codestral and 28.3% for Llama 3.2 3B.
Llama 3.2 3B is cheaper at $0.05 / $0.33 per million input/output tokens, against $0.30 / $0.90 for Codestral.
Codestral accepts more context: 256K tokens versus 131K.
Llama 3.2 3B has downloadable open weights; the other is API-only.

Side by side

Codestral and Llama 3.2 3B specifications
	Codestral	Llama 3.2 3B
Provider	Mistral AI	Meta
Noometry Index	30.6	28.9
Released	2024-05-29	2024-09-24
Weights	Proprietary	Open
Context window	256K	131K
Max output	8K	118K
Input $ / M tokens	$0.30	$0.05
Output $ / M tokens	$0.90	$0.33
Results tracked	7	18

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Too close to call

Codestral: 27.3 (#321), Llama 3.2 3B: 27.6 (#319)

Coding benchmarks
Benchmark	Codestral	Llama 3.2 3B
BigCodeBench Instruct	41.8%	23.4%
BigCodeBench Complete	52.5%	28.3%
Aider Polyglot	11.1%	—
LMArena Coding	—	1098
ALE-Bench	137.78	—
HumanEval+	73.8%	—
MBPP+	61.9%	—

Agentic & Tool Use Not comparable

Codestral: —, Llama 3.2 3B: 20.1 (#143)

Agentic & Tool Use benchmarks
Benchmark	Codestral	Llama 3.2 3B
Berkeley Function Calling Leaderboard	—	21.9%
BALROG	—	10.1%

Reasoning Llama 3.2 3B leads

Codestral: 19.8 (#251), Llama 3.2 3B: 21.0 (#228)

Reasoning benchmarks
Benchmark	Codestral	Llama 3.2 3B
Kagi LLM Benchmark	32.5%	—
LMArena Hard Prompts	—	1095

Math Not comparable

Codestral: —, Llama 3.2 3B: 32.4 (#214)

Math benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Math	—	1126

Knowledge Not comparable

Codestral: —, Llama 3.2 3B: 29.7 (#235)

Knowledge benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Expert	—	1090

Multilingual Not comparable

Codestral: —, Llama 3.2 3B: 26.2 (#281)

Multilingual benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Non-English	—	1019
LMArena Chinese	—	1017
LMArena German	—	1056
LMArena Russian	—	949

Instruction Following Not comparable

Codestral: —, Llama 3.2 3B: 56.0 (#275)

Instruction Following benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Instruction Following	—	1089

Long Context Not comparable

Codestral: —, Llama 3.2 3B: 33.4 (#261)

Long Context benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Longer Query	—	1100

Writing & Preference Not comparable

Codestral: —, Llama 3.2 3B: 24.7 (#307)

Writing & Preference benchmarks
Benchmark	Codestral	Llama 3.2 3B
LMArena Text	—	1110
LMArena Creative Writing	—	1094
EQ-Bench Creative Writing	—	595
LMArena Multi-Turn	—	1105

Frequently asked questions

Is Codestral better than Llama 3.2 3B?

Which is cheaper, Codestral or Llama 3.2 3B?

Llama 3.2 3B is cheaper. It lists at $0.05 per million input tokens and $0.33 per million output tokens; Codestral lists at $0.30 and $0.90.

Is Codestral or Llama 3.2 3B better for coding?

They score almost the same on coding (27.3 vs 27.6); test both on your own repository before choosing.

Which has the bigger context window?

Codestral does, with 256K tokens against 131K.

How many benchmarks do Codestral and Llama 3.2 3B share?

2 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and Llama 3.2 3B has 18.