vs
0:6
overall capability wins (0 ties across 6 scored). Mercury 2 leads overall.
- Reasoning0:4
- Coding0:1
- Pricing3:0
- Specs0:1
Verdict
Mercury 2 leads coding (widest gap: +29.9 on SciCode); Mercury 2 offers ~3.9x larger context (128K vs 33K); LFM2.5-8B-A1B ships open weights (LFM License).
- Mercury 2 leads coding (widest gap: +29.9 on SciCode)
- Mercury 2 offers ~3.9x larger context (128K vs 33K)
- LFM2.5-8B-A1B ships open weights (LFM License)
- LFM2.5-8B-A1B is ~7.1x cheaper on a blended token basis than Mercury 2
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- LFM2.5-8B-A1B
- $0.06
- Mercury 2
- $0.44
LFM2.5-8B-A1B is about 7.3x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Liquid AI
Inception Labs
Family
Other
Other
License
LFM License
Proprietary
Open weights
Yes
No
Release
Jan 20, 2026
Sep 1, 2026
Knowledge cutoff
-
-
API / provider
Liquid AI
Inception Labs
Modalities
text → text
text → text
Specs
Context window
33K
128K
Max output
8K
-
Parameters
8B (1B act.)
—
Pricing
Input $/1M
$0.03
$0.25
Output $/1M
$0.12
$0.75
Blended $/1M (3∶1)
$0.05
$0.38
Speed
tok/s
160
-
TTFT (s)
0.18
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
51.3
77.0
Humanity's Last Exam
6.9
17.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
9.3
21.2
AA-LCR v1.1
—
43.7
CritPt
—
0.8
MMMU-Pro
—
—
IFBench
55.6
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
7.8
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 | LFM2.5-8B-A1B | Mercury 2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Liquid AI | Inception Labs | — |
| Family | Other | Other | — |
| License | LFM License | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Jan 20, 2026 | Sep 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Liquid AI | Inception Labs | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 33K | 128K | B +95K |
| Max output | 8K | - | — |
| Parameters | 8B (1B act.) | — | — |
| Pricing | |||
| Input $/1M | $0.03 | $0.25 | A +$0.22 |
| Output $/1M | $0.12 | $0.75 | A +$0.63 |
| Blended $/1M (3∶1) | $0.05 | $0.38 | A +$0.32 |
| Speed | |||
| tok/s | 160 | - | — |
| TTFT (s) | 0.18 | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 51.3 | 77.0 | B +25.7 pts |
| Humanity's Last Exam | 6.9 | 17.1 | B +10.2 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 9.3 | 21.2 | B +11.9 pts |
| AA-LCR v1.1 | — | 43.7 | — |
| CritPt | — | 0.8 | — |
| MMMU-Pro | — | — | — |
| IFBench | 55.6 | 69.8 | B +14.2 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 | 7.8 | 37.7 | B +29.9 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | 1,167 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
LFM2.5-8B-A1B
- Context
- 33K / 8K out
- Parameters
- 8B (1B act.)
- Price
- $0.03 / $0.12
- Speed
- 160 tok/s · 0.18s TTFT
- Modalities
- text → text
- License
- LFM License
Liquid Foundation Model sparse MoE — efficient open edge/chat model from Liquid AI.
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.