vs
2:5
overall capability wins (0 ties across 7 scored). Mercury 2 leads overall.
- Reasoning1:4
- Coding0:1
- Specs1:0
Verdict
Mercury 2 leads coding (widest gap: +12.3 on TermBench); K2 Think V2 offers larger context (131K vs 128K); K2 Think V2 ships open weights (Apache 2.0).
- Mercury 2 leads coding (widest gap: +12.3 on TermBench)
- K2 Think V2 offers larger context (131K vs 128K)
- K2 Think V2 ships open weights (Apache 2.0)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- K2 Think V2
- -
- Mercury 2
- $0.44
Price per 1M tokens
Full comparison
Organization
MBZUAI / G42 / LLM360
Inception Labs
Family
Other
Other
License
Apache 2.0
Proprietary
Open weights
Yes
No
Release
Jan 27, 2026
Sep 1, 2026
Knowledge cutoff
-
-
API / provider
No primary API price listed
Inception Labs
Modalities
text → text
text → text
Specs
Context window
131K
128K
Max output
-
-
Parameters
70B
—
Pricing
Input $/1M
—
$0.25
Output $/1M
—
$0.75
Blended $/1M (3∶1)
-
$0.38
Speed
tok/s
-
-
TTFT (s)
-
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
71.3
77.0
Humanity's Last Exam
10.1
17.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
18.0
21.2
AA-LCR v1.1
57.0
43.7
CritPt
—
0.8
MMMU-Pro
—
—
IFBench
62.8
69.8
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
15.0
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 | K2 Think V2 | Mercury 2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | MBZUAI / G42 / LLM360 | Inception Labs | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Jan 27, 2026 | Sep 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | Inception Labs | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 131K | 128K | A +3K |
| Max output | - | - | — |
| Parameters | 70B | — | — |
| Pricing | |||
| Input $/1M | — | $0.25 | — |
| Output $/1M | — | $0.75 | — |
| Blended $/1M (3∶1) | - | $0.38 | — |
| Speed | |||
| tok/s | - | - | — |
| TTFT (s) | - | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 71.3 | 77.0 | B +5.7 pts |
| Humanity's Last Exam | 10.1 | 17.1 | B +7.0 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 18.0 | 21.2 | B +3.2 pts |
| AA-LCR v1.1 | 57.0 | 43.7 | A +13.3 pts |
| CritPt | — | 0.8 | — |
| MMMU-Pro | — | — | — |
| IFBench | 62.8 | 69.8 | B +7.0 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 15.0 | 27.3 | B +12.3 pts |
| 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 | — | — | — |
K2 Think V2
- Context
- 131K
- Parameters
- 70B
- Price
- —
- Speed
- —
- Modalities
- text → text
- License
- Apache 2.0
MBZUAI's Institute of Foundation Models, with G42 and LLM360, fully open 70B dense reasoning model built end-to-end as a 'sovereign' successor to the original K2-Think; no hosted API yet.
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.