vs
0:5
overall capability wins (0 ties across 5 scored). Mercury 2 leads overall.
- Reasoning0:4
- Pricing0:3
- Specs0:1
Verdict
Mercury 2 edges reasoning & knowledge; Mercury 2 offers ~2.0x larger context (128K vs 66K); Apertus 70B Instruct ships open weights (Apache 2.0).
- Mercury 2 edges reasoning & knowledge
- Mercury 2 offers ~2.0x larger context (128K vs 66K)
- Apertus 70B Instruct ships open weights (Apache 2.0)
- Mercury 2 is ~3.6x cheaper on a blended token basis than Apertus 70B Instruct
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Apertus 70B Instruct
- $1.55
- Mercury 2
- $0.44
Mercury 2 is about 3.5x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Swiss AI Initiative
Inception Labs
Family
Other
Other
License
Apache 2.0
Proprietary
Open weights
Yes
No
Release
Sep 2, 2025
Sep 1, 2026
Knowledge cutoff
-
-
API / provider
Public AI
Inception Labs
Modalities
text → text
text → text
Specs
Context window
66K
128K
Max output
-
-
Parameters
70B
—
Pricing
Input $/1M
$0.82
$0.25
Output $/1M
$2.92
$0.75
Blended $/1M (3∶1)
$1.34
$0.38
Speed
tok/s
-
-
TTFT (s)
-
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
27.2
77.0
Humanity's Last Exam
5.5
17.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
11.6
21.2
AA-LCR v1.1
—
43.7
CritPt
—
0.8
MMMU-Pro
—
—
IFBench
25.9
69.8
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
27.3
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
—
37.7
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
1,167
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Apertus 70B Instruct | Mercury 2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Swiss AI Initiative | Inception Labs | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Sep 2, 2025 | Sep 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Public AI | Inception Labs | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 66K | 128K | B +62K |
| Max output | - | - | — |
| Parameters | 70B | — | — |
| Pricing | |||
| Input $/1M | $0.82 | $0.25 | B +$0.57 |
| Output $/1M | $2.92 | $0.75 | B +$2.17 |
| Blended $/1M (3∶1) | $1.34 | $0.38 | B +$0.97 |
| Speed | |||
| tok/s | - | - | — |
| TTFT (s) | - | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 27.2 | 77.0 | B +49.8 pts |
| Humanity's Last Exam | 5.5 | 17.1 | B +11.6 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 11.6 | 21.2 | B +9.6 pts |
| AA-LCR v1.1 | — | 43.7 | — |
| CritPt | — | 0.8 | — |
| MMMU-Pro | — | — | — |
| IFBench | 25.9 | 69.8 | B +43.9 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | 27.3 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | — | 37.7 | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | 1,167 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Apertus 70B Instruct
- Context
- 66K
- Parameters
- 70B
- Price
- $0.82 / $2.92
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
Flagship of the Swiss AI Initiative's fully open Apertus family — Apache 2.0 weights, data, and training recipe with broad multilingual coverage including Swiss German and Romansh.
Mercury 2
- Context
- 128K
- Parameters
- —
- Price
- $0.25 / $0.75
- Speed
- —
- Modalities
- text → text
- License
- Proprietary
Inception Labs' diffusion-based (non-autoregressive) reasoning LLM, API-only with a 128K context window and high-throughput inference on Blackwell GPUs.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.