vs
0:8
overall capability wins (0 ties across 8 scored). Mistral Small 4 leads overall.
- Reasoning0:5
- Coding0:1
- Pricing3:0
- Speed0:2
- Specs0:2
Verdict
Mistral Small 4 leads coding (widest gap: +14.5 on SciCode); Mistral Small 4 offers ~2.0x larger context (256K vs 128K); Mistral Small 4 allows ~7.8x more max output (64K vs 8K).
- Mistral Small 4 leads coding (widest gap: +14.5 on SciCode)
- Mistral Small 4 offers ~2.0x larger context (256K vs 128K)
- Mistral Small 4 allows ~7.8x more max output (64K vs 8K)
- Devstral Small is ~2.1x cheaper on a blended token basis than Mistral Small 4
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Devstral Small
- $0.14
- Mistral Small 4
- $0.30
Devstral Small is about 2.1x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Mistral
Mistral
Family
Mistral
Mistral
License
Apache 2.0
Apache 2.0
Open weights
Yes
Yes
Release
May 21, 2025
Mar 1, 2026
Knowledge cutoff
-
-
API / provider
Mistral
Mistral
Modalities
text → text
text, image → text
Specs
Context window
128K
256K
Max output
8K
64K
Parameters
24B
—
Pricing
Input $/1M
$0.07
$0.15
Output $/1M
$0.28
$0.60
Blended $/1M (3∶1)
$0.12
$0.26
Speed
tok/s
100
140
TTFT (s)
0.28
0.2
Reasoning
MMLU-Pro
—
—
GPQA Diamond
41.4
76.9
Humanity's Last Exam
3.8
9.9
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
14.7
21.7
AA-LCR v1.1
18.7
49.7
CritPt
—
0.3
MMMU-Pro
—
56.8
IFBench
34.6
48.2
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
46.8
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
21.0
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
24.3
38.8
CursorBench
—
—
SWE-Rebench
14.6
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Devstral Small | Mistral Small 4 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Mistral | Mistral | — |
| Family | Mistral | Mistral | — |
| License | Apache 2.0 | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | May 21, 2025 | Mar 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Mistral | Mistral | — |
| Modalities | text → text | text, image → text | — |
| Specs | |||
| Context window | 128K | 256K | B +128K |
| Max output | 8K | 64K | B +56K |
| Parameters | 24B | — | — |
| Pricing | |||
| Input $/1M | $0.07 | $0.15 | A +$0.08 |
| Output $/1M | $0.28 | $0.60 | A +$0.32 |
| Blended $/1M (3∶1) | $0.12 | $0.26 | A +$0.14 |
| Speed | |||
| tok/s | 100 | 140 | B +40 |
| TTFT (s) | 0.28 | 0.2 | B +0.08 |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 41.4 | 76.9 | B +35.5 pts |
| Humanity's Last Exam | 3.8 | 9.9 | B +6.1 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 14.7 | 21.7 | B +7.0 pts |
| AA-LCR v1.1 | 18.7 | 49.7 | B +31.0 pts |
| CritPt | — | 0.3 | — |
| MMMU-Pro | — | 56.8 | — |
| IFBench | 34.6 | 48.2 | B +13.6 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | 46.8 | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | 21.0 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 24.3 | 38.8 | B +14.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | 14.6 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Devstral Small
- Context
- 128K / 8K out
- Parameters
- 24B
- Price
- $0.07 / $0.28
- Speed
- 100 tok/s · 0.28s TTFT
- Modalities
- text → text
- License
- Apache 2.0
Open coding agent model from Mistral — lighter sibling to Devstral 2 for IDE workflows.
Mistral Small 4
- Context
- 256K / 64K out
- Parameters
- —
- Price
- $0.15 / $0.60
- Speed
- 140 tok/s · 0.2s TTFT
- Modalities
- text, image → text
- License
- Apache 2.0
Hybrid instruct/reasoning/coding Small model under Apache 2.0 - Mistral's high-throughput production default.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.