A · open · Mistral

Mistral Medium 3.5

Mistral

B · open · Mistral

Mistral Small 3.2

Mistral

Replace A

Replace B

12:0

overall capability wins (0 ties across 12 scored). Mistral Medium 3.5 leads overall.

Reasoning6:0
Coding2:0
Tool use1:0
Composite indices2:0
Pricing0:3
Speed0:2
Specs3:0

Verdict

Mistral Medium 3.5 leads coding (widest gap: +45.0 on TermBench); Mistral Medium 3.5 offers ~2.0x larger context (256K vs 128K); Mistral Medium 3.5 allows ~2.0x more max output (64K vs 33K).

Mistral Medium 3.5 leads coding (widest gap: +45.0 on TermBench)
Mistral Medium 3.5 offers ~2.0x larger context (256K vs 128K)
Mistral Medium 3.5 allows ~2.0x more max output (64K vs 33K)
Mistral Small 3.2 is ~20.0x cheaper on a blended token basis than Mistral Medium 3.5

Cost for 1M in + 250K out

Illustrative chat workload at primary-provider list prices.

Mistral Medium 3.5: $3.38
Mistral Small 3.2: $0.17

Mistral Small 3.2 is about 19.3x cheaper on this workload.

Price per 1M tokens

Full comparison

Organization

Mistral

Family

Mistral

License

Modified MIT

Apache 2.0

Open weights

Yes

Release

Apr 22, 2026

Jun 20, 2025

Knowledge cutoff

API / provider

Mistral

Modalities

text, image → text

Specs

Context window

256K

128K

A +128K

Max output

64K

33K

A +31K

Parameters

128B

24B

A +104

Pricing

Input $/1M

$1.50

$0.10

B +$1.40

Output $/1M

$7.50

$0.30

B +$7.20

Blended $/1M (3∶1)

$0.15

B +$2.85

Speed

tok/s

140

B +60

TTFT (s)

0.4

0.2

B +0.20

Reasoning

MMLU-Pro

—

69.1

—

GPQA Diamond

74.8

50.5

A +24.3 pts

Humanity's Last Exam

13.8

4.3

A +9.5 pts

AIME 2025

—

27.0

—

MATH-500

—

88.3

—

Humanity's Last Exam (with tools)

—

AA-Omniscience Accuracy

24.7

14.8

A +9.9 pts

AA-LCR v1.1

69.3

20.3

A +49.0 pts

CritPt

—

MMMU-Pro

64.9

48.0

A +16.9 pts

IFBench

68.8

33.5

A +35.3 pts

Chartography

—

Chartography (With Tools)

—

Coding

SWE-bench Verified

—

SWE-bench Pro

—

SWE-bench Multilingual

—

LiveCodeBench

—

27.5

—

Terminal-Bench 2.1

50.6

5.6

A +45.0 pts

Aider Polyglot

—

Terminal-Bench 3

—

BigCodeBench

—

SciCode

40.2

28.6

A +11.6 pts

CursorBench

—

SWE-Rebench

—

NL2Repo-Bench

—

DeepSWE

—

WebDev Arena

1,265

—

Terminal-Bench 4.0

—

CursorBench 4.0

—

FrontierCode v1.1 (Main)

—

Arena

LMArena Elo

1,426

—

Metric	Mistral Medium 3.5	Mistral Small 3.2	Delta
Identity
Organization	Mistral	Mistral	—
Family	Mistral	Mistral	—
License	Modified MIT	Apache 2.0	—
Open weights	Yes	Yes	—
Release	Apr 22, 2026	Jun 20, 2025	—
Knowledge cutoff	-	-	—
API / provider	Mistral	Mistral	—
Modalities	text, image → text	text, image → text	—
Specs
Context window	256K	128K	A +128K
Max output	64K	33K	A +31K
Parameters	128B	24B	A +104
Pricing
Input $/1M	$1.50	$0.10	B +$1.40
Output $/1M	$7.50	$0.30	B +$7.20
Blended $/1M (3∶1)	$3	$0.15	B +$2.85
Speed
tok/s	80	140	B +60
TTFT (s)	0.4	0.2	B +0.20
Reasoning
MMLU-Pro	—	69.1	—
GPQA Diamond	74.8	50.5	A +24.3 pts
Humanity's Last Exam	13.8	4.3	A +9.5 pts
AIME 2025	—	27.0	—
MATH-500	—	88.3	—
Humanity's Last Exam (with tools)	—	—	—
AA-Omniscience Accuracy	24.7	14.8	A +9.9 pts
AA-LCR v1.1	69.3	20.3	A +49.0 pts
CritPt	—	—	—
MMMU-Pro	64.9	48.0	A +16.9 pts
IFBench	68.8	33.5	A +35.3 pts
Chartography	—	—	—
Chartography (With Tools)	—	—	—
Coding
SWE-bench Verified	—	—	—
SWE-bench Pro	—	—	—
SWE-bench Multilingual	—	—	—
LiveCodeBench	—	27.5	—
Terminal-Bench 2.1	50.6	5.6	A +45.0 pts
Aider Polyglot	—	—	—
Terminal-Bench 3	—	—	—
BigCodeBench	—	—	—
SciCode	40.2	28.6	A +11.6 pts
CursorBench	—	—	—
SWE-Rebench	—	—	—
NL2Repo-Bench	—	—	—
DeepSWE	—	—	—
WebDev Arena	1,265	—	—
Terminal-Bench 4.0	—	—	—
CursorBench 4.0	—	—	—
FrontierCode v1.1 (Main)	—	—	—
Arena
LMArena Elo	1,426	—	—