vs
6:1
overall capability wins (0 ties across 7 scored). Devstral 2 leads overall.
- Reasoning3:1
- Pricing0:3
- Speed0:2
- Specs3:0
Verdict
Devstral 2 edges reasoning & knowledge; Devstral 2 offers ~31.3x larger context (256K vs 8K); Devstral 2 allows ~7.8x more max output (64K vs 8K).
- Devstral 2 edges reasoning & knowledge
- Devstral 2 offers ~31.3x larger context (256K vs 8K)
- Devstral 2 allows ~7.8x more max output (64K vs 8K)
- Mistral 7B is ~13.3x cheaper on a blended token basis than Devstral 2
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Devstral 2
- $0.90
- Mistral 7B
- $0.07
Mistral 7B is about 12.0x 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
Dec 1, 2025
Sep 27, 2023
Knowledge cutoff
-
-
API / provider
Mistral
Together / Fireworks (ref.)
Modalities
text → text
text → text
Specs
Context window
256K
8K
Max output
64K
8K
Parameters
123B
7B
Pricing
Input $/1M
$0.40
$0.06
Output $/1M
$2
$0.06
Blended $/1M (3∶1)
$0.80
$0.06
Speed
tok/s
85
180
TTFT (s)
0.35
0.15
Reasoning
MMLU-Pro
76.2
—
GPQA Diamond
59.4
17.7
Humanity's Last Exam
3.6
4.6
AIME 2025
36.7
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
20.8
9.0
AA-LCR v1.1
32.3
—
CritPt
—
—
MMMU-Pro
—
—
IFBench
38.1
19.9
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
44.8
—
Terminal-Bench 2.1
30.3
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
19.5
SciCode
32.8
—
CursorBench
—
—
SWE-Rebench
42.0
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,194
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
1,110
| Metric | Devstral 2 | Mistral 7B | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Mistral | Mistral | — |
| Family | Mistral | Mistral | — |
| License | Apache 2.0 | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | Dec 1, 2025 | Sep 27, 2023 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Mistral | Together / Fireworks (ref.) | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 256K | 8K | A +248K |
| Max output | 64K | 8K | A +56K |
| Parameters | 123B | 7B | A +116 |
| Pricing | |||
| Input $/1M | $0.40 | $0.06 | B +$0.34 |
| Output $/1M | $2 | $0.06 | B +$1.94 |
| Blended $/1M (3∶1) | $0.80 | $0.06 | B +$0.74 |
| Speed | |||
| tok/s | 85 | 180 | B +95 |
| TTFT (s) | 0.35 | 0.15 | B +0.20 |
| Reasoning | |||
| MMLU-Pro | 76.2 | — | — |
| GPQA Diamond | 59.4 | 17.7 | A +41.7 pts |
| Humanity's Last Exam | 3.6 | 4.6 | B +1.0 pts |
| AIME 2025 | 36.7 | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 20.8 | 9.0 | A +11.8 pts |
| AA-LCR v1.1 | 32.3 | — | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | 38.1 | 19.9 | A +18.2 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 44.8 | — | — |
| Terminal-Bench 2.1 | 30.3 | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | 19.5 | — |
| SciCode | 32.8 | — | — |
| CursorBench | — | — | — |
| SWE-Rebench | 42.0 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,194 | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | 1,110 | — |
Devstral 2
- Context
- 256K / 64K out
- Parameters
- 123B
- Price
- $0.40 / $2
- Speed
- 85 tok/s · 0.35s TTFT
- Modalities
- text → text
- License
- Apache 2.0
Open-weight agentic coding model for autonomous software engineering (successor path: Medium 3.5).
Mistral 7B
- Context
- 8K / 8K out
- Parameters
- 7B
- Price
- $0.06 / $0.06
- Speed
- 180 tok/s · 0.15s TTFT
- Modalities
- text → text
- License
- Apache 2.0
The 7B that put Mistral on the map — still a default fine-tune and local-chat base.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.