Model comparison

Codestral vs DeepSeek-V3

DeepSeek-V3 is the stronger model overall, scoring 39.5 to 30.6 on the Noometry Index.

Last verified October 11, 2026. 6 shared benchmarks.

Codestral

Mistral AI

30.6

Rank #290 Reported

DeepSeek-V3

DeepSeek

39.5

Rank #166 Confirmed

Summary

They share 6 benchmarks with published results for both. Codestral scores higher in 0 categories and DeepSeek-V3 in 2 categories; one gap is clear of the uncertainty.
The widest gap is in coding, where DeepSeek-V3 leads 42.3 to 27.3.
The biggest single-benchmark swing is Aider Polyglot: 11.1% for Codestral and 55.1% for DeepSeek-V3.
DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $0.30 / $0.90 for Codestral.
Codestral accepts more context: 256K tokens versus 164K.
DeepSeek-V3 has downloadable open weights; the other is API-only.

Side by side

Codestral and DeepSeek-V3 specifications
	Codestral	DeepSeek-V3
Provider	Mistral AI	DeepSeek
Noometry Index	30.6	39.5
Released	2024-05-29	2024-12-26
Weights	Proprietary	Open
Context window	256K	164K
Max output	8K	164K
Input $ / M tokens	$0.30	$0.24
Output $ / M tokens	$0.90	$0.90
Results tracked	7	60

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

Category by category

Coding DeepSeek-V3 leads

Codestral: 27.3 (#321), DeepSeek-V3: 42.3 (#106)

Coding benchmarks
Benchmark	Codestral	DeepSeek-V3
Aider Polyglot	11.1%	55.1%
BigCodeBench Instruct	41.8%	50%
BigCodeBench Complete	52.5%	62.2%
HumanEval+	73.8%	86.6%
MBPP+	61.9%	73%
SciCode	—	35.8%
WeirdML	—	36.1%
LiveBench Coding	—	70.9%
LMArena Coding	—	1368
ALE-Bench	137.78	—

Agentic & Tool Use Not comparable

Codestral: —, DeepSeek-V3: —

Agentic & Tool Use benchmarks
Benchmark	Codestral	DeepSeek-V3
METR Time Horizons	—	49.6%

Reasoning Too close to call

Codestral: 19.8 (#251), DeepSeek-V3: 20.5 (#236)

Reasoning benchmarks
Benchmark	Codestral	DeepSeek-V3
Kagi LLM Benchmark	32.5%	52.3%
SimpleBench	—	27.2%
CritPt	—	0%
LiveBench Reasoning	—	65.8%
LMArena Hard Prompts	—	1365
DTBench	—	64.8%
LiveBench Data Analysis	—	60.9%
LMCA	—	15.5%
BIG-Bench Hard	—	87.5%
Epoch Capabilities Index	—	135.94
ForecastBench	—	59.1
HellaSwag	—	88.9%
LiveBench	—	66.9%
PIQA	—	84.7%
WinoGrande	—	85.2%

Math Not comparable

Codestral: —, DeepSeek-V3: 32.1 (#219)

Math benchmarks
Benchmark	Codestral	DeepSeek-V3
OTIS Mock AIME 2024-2025	—	37.8%
Omni-MATH	—	40.3%
LiveBench Math	—	73.5%
LMArena Math	—	1373
MATH Level 5	—	75.5%
FrontierMath (Feb 2025 set)	—	1.7%

Knowledge Not comparable

Codestral: —, DeepSeek-V3: 37.5 (#155)

Knowledge benchmarks
Benchmark	Codestral	DeepSeek-V3
GPQA Diamond	—	67.6%
MMLU-Pro	—	72.3%
Confabulations	—	26.1%
Vectara Hallucination Rate	—	6.1%
GPQA (HELM)	—	53.8%
LMArena Expert	—	1351
ARC (AI2) Challenge	—	95.3%
MMLU	—	87.2%
TriviaQA	—	82.9%

Multilingual Not comparable

Codestral: —, DeepSeek-V3: 48.5 (#143)

Multilingual benchmarks
Benchmark	Codestral	DeepSeek-V3
LMArena Non-English	—	1358
LMArena Chinese	—	1391
LMArena French	—	1385
LMArena German	—	1374
LMArena Japanese	—	1333
LMArena Korean	—	1319
LMArena Russian	—	1373
LMArena Spanish	—	1358

Instruction Following Not comparable

Codestral: —, DeepSeek-V3: 72.8 (#130)

Instruction Following benchmarks
Benchmark	Codestral	DeepSeek-V3
LiveBench Instruction Following	—	81.5%
IFEval	—	83.2%
LMArena Instruction Following	—	1345

Long Context Not comparable

Codestral: —, DeepSeek-V3: 34.0 (#253)

Long Context benchmarks
Benchmark	Codestral	DeepSeek-V3
Fiction.LiveBench	—	50%
LMArena Longer Query	—	1352

Writing & Preference Not comparable

Codestral: —, DeepSeek-V3: 57.4 (#130)

Writing & Preference benchmarks
Benchmark	Codestral	DeepSeek-V3
LMArena Text	—	1375
LMArena Creative Writing	—	1364
Short-Story Creative Writing	—	77%
EQ-Bench Creative Writing	—	1472
WildBench	—	83%
LMArena Multi-Turn	—	1389
LiveBench Language	—	49.1%

Frequently asked questions

Is Codestral better than DeepSeek-V3?

DeepSeek-V3 is the stronger model overall, scoring 39.5 to 30.6 on the Noometry Index.

Which is cheaper, Codestral or DeepSeek-V3?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Codestral lists at $0.30 and $0.90.

Is Codestral or DeepSeek-V3 better for coding?

DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 27.3 in the Noometry coding category.

Which has the bigger context window?

Codestral does, with 256K tokens against 164K.

How many benchmarks do Codestral and DeepSeek-V3 share?

6 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and DeepSeek-V3 has 60.