vs
0:2
overall capability wins (0 ties across 2 scored). K2 Horizon 375B A23B leads overall.
- Reasoning0:1
- Specs0:1
Verdict
K2 Horizon 375B A23B edges reasoning & knowledge; K2 Horizon 375B A23B offers ~32.0x larger context (524K vs 16K); K2 Horizon 375B A23B ships open weights (Apache 2.0).
- K2 Horizon 375B A23B edges reasoning & knowledge
- K2 Horizon 375B A23B offers ~32.0x larger context (524K vs 16K)
- K2 Horizon 375B A23B ships open weights (Apache 2.0)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- GPT-3.5 Turbo
- $0.88
- K2 Horizon 375B A23B
- -
Price per 1M tokens
Full comparison
Organization
OpenAI
MBZUAI Institute of Foundation Models
Family
Other
Other
License
Proprietary
Apache 2.0
Open weights
No
Yes
Release
Mar 1, 2023
Sep 3, 2026
Knowledge cutoff
-
-
API / provider
OpenAI
No primary API price listed
Modalities
text → text
text → text
Specs
Context window
16K
524K
Max output
4K
-
Parameters
—
375B (23B act.)
Pricing
Input $/1M
$0.50
—
Output $/1M
$1.50
—
Blended $/1M (3∶1)
$0.75
-
Speed
tok/s
120
-
TTFT (s)
0.25
-
Reasoning
MMLU-Pro
46.2
—
GPQA Diamond
29.7
87.3
Humanity's Last Exam
—
32.0
AIME 2025
—
—
MATH-500
44.1
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
18.2
AA-LCR v1.1
—
80.0
CritPt
—
4.6
MMMU-Pro
—
—
IFBench
—
—
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
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,225
—
| Metric | GPT-3.5 Turbo | K2 Horizon 375B A23B | Delta |
|---|---|---|---|
| Identity | |||
| Organization | OpenAI | MBZUAI Institute of Foundation Models | — |
| Family | Other | Other | — |
| License | Proprietary | Apache 2.0 | — |
| Open weights | No | Yes | — |
| Release | Mar 1, 2023 | Sep 3, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | OpenAI | No primary API price listed | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 16K | 524K | B +508K |
| Max output | 4K | - | — |
| Parameters | — | 375B (23B act.) | — |
| Pricing | |||
| Input $/1M | $0.50 | — | — |
| Output $/1M | $1.50 | — | — |
| Blended $/1M (3∶1) | $0.75 | - | — |
| Speed | |||
| tok/s | 120 | - | — |
| TTFT (s) | 0.25 | - | — |
| Reasoning | |||
| MMLU-Pro | 46.2 | — | — |
| GPQA Diamond | 29.7 | 87.3 | B +57.6 pts |
| Humanity's Last Exam | — | 32.0 | — |
| AIME 2025 | — | — | — |
| MATH-500 | 44.1 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | 18.2 | — |
| AA-LCR v1.1 | — | 80.0 | — |
| CritPt | — | 4.6 | — |
| MMMU-Pro | — | — | — |
| IFBench | — | — | — |
| 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 | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,225 | — | — |
GPT-3.5 Turbo
- Context
- 16K / 4K out
- Parameters
- —
- Price
- $0.50 / $1.50
- Speed
- 120 tok/s · 0.25s TTFT
- Modalities
- text → text
- License
- Proprietary
Classic ChatGPT-era budget model — still referenced for cheap classification and simple chat.
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.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.