vs
5:0
overall capability wins (0 ties across 5 scored). Agnes-3.0-Flash Preview leads overall.
- Reasoning3:0
- Specs2:0
Verdict
Agnes-3.0-Flash Preview edges reasoning & knowledge; Agnes-3.0-Flash Preview offers ~4.0x larger context (262K vs 66K).
- Agnes-3.0-Flash Preview edges reasoning & knowledge
- Agnes-3.0-Flash Preview offers ~4.0x larger context (262K vs 66K)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Agnes-3.0-Flash Preview
- -
- Apertus 8B Instruct
- $0.15
Price per 1M tokens
Full comparison
Organization
Agnes AI
Swiss AI Initiative
Family
Other
Other
License
Apache 2.0
Apache 2.0
Open weights
Yes
Yes
Release
Sep 13, 2026
Sep 2, 2025
Knowledge cutoff
-
-
API / provider
No primary API price listed
Public AI
Modalities
text, image, video → text
text → text
Specs
Context window
262K
66K
Max output
-
-
Parameters
33B
8B
Pricing
Input $/1M
—
$0.10
Output $/1M
—
$0.20
Blended $/1M (3∶1)
-
$0.13
Speed
tok/s
-
-
TTFT (s)
-
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
85.0
25.6
Humanity's Last Exam
—
5.0
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
23.0
9.7
AA-LCR v1.1
68.3
—
CritPt
—
—
MMMU-Pro
—
—
IFBench
74.2
22.4
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
38.1
—
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Agnes-3.0-Flash Preview | Apertus 8B Instruct | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Agnes AI | Swiss AI Initiative | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Apache 2.0 | — |
| Open weights | Yes | Yes | — |
| Release | Sep 13, 2026 | Sep 2, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | Public AI | — |
| Modalities | text, image, video → text | text → text | — |
| Specs | |||
| Context window | 262K | 66K | A +197K |
| Max output | - | - | — |
| Parameters | 33B | 8B | A +25 |
| Pricing | |||
| Input $/1M | — | $0.10 | — |
| Output $/1M | — | $0.20 | — |
| Blended $/1M (3∶1) | - | $0.13 | — |
| Speed | |||
| tok/s | - | - | — |
| TTFT (s) | - | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 85.0 | 25.6 | A +59.4 pts |
| Humanity's Last Exam | — | 5.0 | — |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 23.0 | 9.7 | A +13.3 pts |
| AA-LCR v1.1 | 68.3 | — | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | 74.2 | 22.4 | A +51.8 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 38.1 | — | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Agnes-3.0-Flash Preview
- Context
- 262K
- Parameters
- 33B
- Price
- —
- Speed
- —
- Modalities
- text, image, video → text
- License
- Apache 2.0
Agnes AI's (Sapiens AI, Singapore) first open-weight release: a 33B-parameter hybrid-attention model (54 gated delta-rule recurrent layers alternating with 18 global-attention layers, 3:1) under Apache 2.0, with a 262K-token context window and text/image/video input. This open-weight Preview checkpoint is distinct from Agnes AI's production "Agnes 3.0 Flash" API model, which runs a different checkpoint and is not open-weight; the two should not be conflated.
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.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.