vs
4:5
overall capability wins (0 ties across 9 scored). MiniMax M3 leads overall.
- Reasoning2:3
- Coding1:1
- Tool use1:0
- Composite indices2:0
- Specs0:1
Verdict
MiniMax M3 edges reasoning & knowledge; MiniMax M3 offers ~1.9x larger context (1M vs 524K).
- MiniMax M3 edges reasoning & knowledge
- MiniMax M3 offers ~1.9x larger context (1M vs 524K)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- K2 Horizon 375B A23B
- -
- MiniMax M3
- $0.60
Price per 1M tokens
Full comparison
Organization
MBZUAI Institute of Foundation Models
MiniMax
Family
Other
Other
License
Apache 2.0
MiniMax Community
Open weights
Yes
Yes
Release
Sep 3, 2026
Jun 1, 2026
Knowledge cutoff
-
-
API / provider
No primary API price listed
MiniMax
Modalities
text → text
text, image, video → text
Specs
Context window
524K
1M
Max output
-
128K
Parameters
375B (23B act.)
—
Pricing
Input $/1M
—
$0.30
Output $/1M
—
$1.20
Blended $/1M (3∶1)
-
$0.52
Speed
tok/s
-
80
TTFT (s)
-
0.35
Reasoning
MMLU-Pro
—
—
GPQA Diamond
87.3
92.9
Humanity's Last Exam
32.0
39.0
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
18.2
16.7
AA-LCR v1.1
80.0
83.0
CritPt
4.6
3.7
MMMU-Pro
—
78.6
IFBench
—
82.9
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
59.0
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
71.9
65.2
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
42.9
47.1
CursorBench
—
—
SWE-Rebench
—
46.4
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
1,487
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
1,441
| Metric | K2 Horizon 375B A23B | MiniMax M3 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | MBZUAI Institute of Foundation Models | MiniMax | — |
| Family | Other | Other | — |
| License | Apache 2.0 | MiniMax Community | — |
| Open weights | Yes | Yes | — |
| Release | Sep 3, 2026 | Jun 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | MiniMax | — |
| Modalities | text → text | text, image, video → text | — |
| Specs | |||
| Context window | 524K | 1M | B +476K |
| Max output | - | 128K | — |
| Parameters | 375B (23B act.) | — | — |
| Pricing | |||
| Input $/1M | — | $0.30 | — |
| Output $/1M | — | $1.20 | — |
| Blended $/1M (3∶1) | - | $0.52 | — |
| Speed | |||
| tok/s | - | 80 | — |
| TTFT (s) | - | 0.35 | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 87.3 | 92.9 | B +5.6 pts |
| Humanity's Last Exam | 32.0 | 39.0 | B +7.0 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 18.2 | 16.7 | A +1.5 pts |
| AA-LCR v1.1 | 80.0 | 83.0 | B +3.0 pts |
| CritPt | 4.6 | 3.7 | A +0.9 pts |
| MMMU-Pro | — | 78.6 | — |
| IFBench | — | 82.9 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | 59.0 | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 71.9 | 65.2 | A +6.7 pts |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 42.9 | 47.1 | B +4.2 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | 46.4 | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | 1,487 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | 1,441 | — |
K2 Horizon 375B A23B
- Context
- 524K
- Parameters
- 375B (23B act.)
- Price
- —
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
Flagship model (375B MoE / 23B active) in the Institute of Foundation Models' K2 Horizon fleet, released fully open — weights, code, training data, and methodology — under Apache 2.0.
MiniMax M3
- Context
- 1M / 128K out
- Parameters
- —
- Price
- $0.30 / $1.20
- Speed
- 80 tok/s · 0.35s TTFT
- Modalities
- text, image, video → text
- License
- MiniMax Community
MiniMax June 2026 open-weight flagship — frontier coding/agentic scores with MSA 1M context and native multimodality.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.