vs
0:3
overall capability wins (0 ties across 3 scored). Hy4 Preview leads overall.
- Reasoning0:1
- Pricing3:0
- Specs0:2
Verdict
Hy4 Preview edges reasoning & knowledge; Hy4 Preview offers ~15.3x larger context (1M vs 66K); Apertus 8B Instruct is ~10.0x cheaper on a blended token basis than Hy4 Preview.
- Hy4 Preview edges reasoning & knowledge
- Hy4 Preview offers ~15.3x larger context (1M vs 66K)
- Apertus 8B Instruct is ~10.0x cheaper on a blended token basis than Hy4 Preview
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Apertus 8B Instruct
- $0.15
- Hy4 Preview
- $1.46
Apertus 8B Instruct is about 9.7x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Swiss AI Initiative
Tencent
Family
Other
Other
License
Apache 2.0
Apache 2.0
Open weights
Yes
Yes
Release
Sep 2, 2025
Aug 28, 2026
Knowledge cutoff
-
-
API / provider
Public AI
Tencent Cloud TokenHub
Modalities
text → text
text → text
Specs
Context window
66K
1M
Max output
-
-
Parameters
8B
770B (49B act.)
Pricing
Input $/1M
$0.10
$0.83
Output $/1M
$0.20
$2.50
Blended $/1M (3∶1)
$0.13
$1.25
Speed
tok/s
-
-
TTFT (s)
-
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
25.6
92.3
Humanity's Last Exam
5.0
—
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
9.7
—
AA-LCR v1.1
—
—
CritPt
—
—
MMMU-Pro
—
—
IFBench
22.4
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
65.7
SWE-bench Multilingual
—
82.9
LiveCodeBench
—
—
Terminal-Bench 2.1
—
85.4
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
—
—
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
64.3
WebDev Arena
—
1,624
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Apertus 8B Instruct | Hy4 Preview | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Swiss AI Initiative | Tencent | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | Sep 2, 2025 | Aug 28, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Public AI | Tencent Cloud TokenHub | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 66K | 1M | B +934K |
| Max output | - | - | — |
| Parameters | 8B | 770B (49B act.) | B +762 |
| Pricing | |||
| Input $/1M | $0.10 | $0.83 | A +$0.73 |
| Output $/1M | $0.20 | $2.50 | A +$2.30 |
| Blended $/1M (3∶1) | $0.13 | $1.25 | A +$1.13 |
| Speed | |||
| tok/s | - | - | — |
| TTFT (s) | - | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 25.6 | 92.3 | B +66.7 pts |
| Humanity's Last Exam | 5.0 | — | — |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 9.7 | — | — |
| AA-LCR v1.1 | — | — | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | 22.4 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | 65.7 | — |
| SWE-bench Multilingual | — | 82.9 | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | 85.4 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | — | — | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | 64.3 | — |
| WebDev Arena | — | 1,624 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Apertus 8B Instruct
- Context
- 66K
- Parameters
- 8B
- Price
- $0.10 / $0.20
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
Fully open, transparent Swiss LLM from EPFL, ETH Zurich, and CSCS — Apache 2.0 weights, training data, and methodology, covering 1,000+ languages.
Hy4 Preview
- Context
- 1M
- Parameters
- 770B (49B act.)
- Price
- $0.83 / $2.50
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
Tencent Hy Team's Aug 28, 2026 open-weight flagship (770B MoE / 49B active, Apache 2.0) with a 1M context window and default high reasoning effort. Text-only instruct weights on Hugging Face; hosted API at $0.834/$2.501 per 1M tokens via TokenHub and OpenRouter.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.