vs
1:0
overall capability wins (0 ties across 1 scored). Mercury 2 leads overall.
- Pricing3:0
- Specs1:0
Verdict
Mercury 2 offers ~8.0x larger context (128K vs 16K); Mercury 2 is ~7.0x cheaper on a blended token basis than Step-2.
- Mercury 2 offers ~8.0x larger context (128K vs 16K)
- Mercury 2 is ~7.0x cheaper on a blended token basis than Step-2
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Mercury 2
- $0.44
- Step-2
- $3
Mercury 2 is about 6.9x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Inception Labs
StepFun
Family
Other
Other
License
Proprietary
Proprietary
Open weights
No
No
Release
Sep 1, 2026
Jan 15, 2025
Knowledge cutoff
-
-
API / provider
Inception Labs
StepFun
Modalities
text → text
text → text
Specs
Context window
128K
16K
Max output
-
8K
Parameters
—
—
Pricing
Input $/1M
$0.25
$1.50
Output $/1M
$0.75
$6
Blended $/1M (3∶1)
$0.38
$2.63
Speed
tok/s
-
50
TTFT (s)
-
0.6
Reasoning
MMLU-Pro
—
—
GPQA Diamond
77.0
—
Humanity's Last Exam
17.1
—
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
21.2
—
AA-LCR v1.1
43.7
—
CritPt
0.8
—
MMMU-Pro
—
—
IFBench
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 | Mercury 2 | Step-2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Inception Labs | StepFun | — |
| Family | Other | Other | — |
| License | Proprietary | Proprietary | — |
| Open weights | No | No | — |
| Release | Sep 1, 2026 | Jan 15, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Inception Labs | StepFun | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 128K | 16K | A +112K |
| Max output | - | 8K | — |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $0.25 | $1.50 | A +$1.25 |
| Output $/1M | $0.75 | $6 | A +$5.25 |
| Blended $/1M (3∶1) | $0.38 | $2.63 | A +$2.25 |
| Speed | |||
| tok/s | - | 50 | — |
| TTFT (s) | - | 0.6 | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 77.0 | — | — |
| Humanity's Last Exam | 17.1 | — | — |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 21.2 | — | — |
| AA-LCR v1.1 | 43.7 | — | — |
| CritPt | 0.8 | — | — |
| MMMU-Pro | — | — | — |
| IFBench | 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 | — | — | — |
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.
Step-2
- Context
- 16K / 8K out
- Parameters
- —
- Price
- $1.50 / $6
- Speed
- 50 tok/s · 0.6s TTFT
- Modalities
- text → text
- License
- Proprietary
StepFun frontier reasoning model that broke into Chinese and global arenas in early 2025.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.
These two models have no overlapping published benchmarks in our dataset. Compare specs and pricing instead.