vs
6:1
overall capability wins (0 ties across 7 scored). K2 Horizon 375B A23B leads overall.
- Reasoning4:1
- Coding1:0
- Specs1:0
Verdict
K2 Horizon 375B A23B leads coding (widest gap: +2.5 on SciCode); K2 Horizon 375B A23B offers ~2.0x larger context (524K vs 256K); K2 Horizon 375B A23B ships open weights (Apache 2.0).
- K2 Horizon 375B A23B leads coding (widest gap: +2.5 on SciCode)
- K2 Horizon 375B A23B offers ~2.0x larger context (524K vs 256K)
- K2 Horizon 375B A23B ships open weights (Apache 2.0)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- K2 Horizon 375B A23B
- -
- Step-3.5 Flash
- $0.16
Price per 1M tokens
Full comparison
Organization
MBZUAI Institute of Foundation Models
StepFun
Family
Other
Other
License
Apache 2.0
Proprietary
Open weights
Yes
No
Release
Sep 3, 2026
Feb 18, 2026
Knowledge cutoff
-
-
API / provider
No primary API price listed
StepFun
Modalities
text → text
text, image → text
Specs
Context window
524K
256K
Max output
-
16K
Parameters
375B (23B act.)
—
Pricing
Input $/1M
—
$0.09
Output $/1M
—
$0.30
Blended $/1M (3∶1)
-
$0.14
Speed
tok/s
-
120
TTFT (s)
-
0.25
Reasoning
MMLU-Pro
—
—
GPQA Diamond
87.3
83.1
Humanity's Last Exam
32.0
21.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
18.2
23.6
AA-LCR v1.1
80.0
50.0
CritPt
4.6
2.5
MMMU-Pro
—
—
IFBench
—
64.6
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
71.9
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
42.9
40.4
CursorBench
—
—
SWE-Rebench
—
59.6
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
1,394
| Metric | K2 Horizon 375B A23B | Step-3.5 Flash | Delta |
|---|---|---|---|
| Identity | |||
| Organization | MBZUAI Institute of Foundation Models | StepFun | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Sep 3, 2026 | Feb 18, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | StepFun | — |
| Modalities | text → text | text, image → text | — |
| Specs | |||
| Context window | 524K | 256K | A +268K |
| Max output | - | 16K | — |
| Parameters | 375B (23B act.) | — | — |
| Pricing | |||
| Input $/1M | — | $0.09 | — |
| Output $/1M | — | $0.30 | — |
| Blended $/1M (3∶1) | - | $0.14 | — |
| Speed | |||
| tok/s | - | 120 | — |
| TTFT (s) | - | 0.25 | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 87.3 | 83.1 | A +4.2 pts |
| Humanity's Last Exam | 32.0 | 21.1 | A +10.9 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 18.2 | 23.6 | B +5.4 pts |
| AA-LCR v1.1 | 80.0 | 50.0 | A +30.0 pts |
| CritPt | 4.6 | 2.5 | A +2.1 pts |
| MMMU-Pro | — | — | — |
| IFBench | — | 64.6 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 71.9 | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 42.9 | 40.4 | A +2.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | 59.6 | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | 1,394 | — |
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.
Step-3.5 Flash
- Context
- 256K / 16K out
- Parameters
- —
- Price
- $0.09 / $0.30
- Speed
- 120 tok/s · 0.25s TTFT
- Modalities
- text, image → text
- License
- Proprietary
StepFun's fast multimodal API tier — competitive STEM scores at Flash pricing.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.