vs
1:0
overall capability wins (0 ties across 1 scored). InternLM2.5 20B leads overall.
- Pricing2:1
- Specs1:0
Verdict
InternLM2.5 20B offers ~7.8x larger context (1M vs 128K); InternLM2.5 20B ships open weights (Apache 2.0); InternLM2.5 20B is ~1.3x cheaper on a blended token basis than Mercury 2.
- InternLM2.5 20B offers ~7.8x larger context (1M vs 128K)
- InternLM2.5 20B ships open weights (Apache 2.0)
- InternLM2.5 20B is ~1.3x cheaper on a blended token basis than Mercury 2
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- InternLM2.5 20B
- $0.38
- Mercury 2
- $0.44
InternLM2.5 20B is about 1.2x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Shanghai AI Lab
Inception Labs
Family
Other
Other
License
Apache 2.0
Proprietary
Open weights
Yes
No
Release
Aug 1, 2024
Sep 1, 2026
Knowledge cutoff
-
-
API / provider
Together / Fireworks (ref.)
Inception Labs
Modalities
text → text
text → text
Specs
Context window
1M
128K
Max output
8K
-
Parameters
20B
—
Pricing
Input $/1M
$0.30
$0.25
Output $/1M
$0.30
$0.75
Blended $/1M (3∶1)
$0.30
$0.38
Speed
tok/s
70
-
TTFT (s)
0.4
-
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 | InternLM2.5 20B | Mercury 2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Shanghai AI Lab | Inception Labs | — |
| Family | Other | Other | — |
| License | Apache 2.0 | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Aug 1, 2024 | Sep 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Together / Fireworks (ref.) | Inception Labs | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 1M | 128K | A +872K |
| Max output | 8K | - | — |
| Parameters | 20B | — | — |
| Pricing | |||
| Input $/1M | $0.30 | $0.25 | B +$0.05 |
| Output $/1M | $0.30 | $0.75 | A +$0.45 |
| Blended $/1M (3∶1) | $0.30 | $0.38 | A +$0.08 |
| Speed | |||
| tok/s | 70 | - | — |
| TTFT (s) | 0.4 | - | — |
| 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 | — | — | — |
InternLM2.5 20B
- Context
- 1M / 8K out
- Parameters
- 20B
- Price
- $0.30 / $0.30
- Speed
- 70 tok/s · 0.4s TTFT
- Modalities
- text → text
- License
- Apache 2.0
Open InternLM2.5 chat with up to 1M context — strong Chinese RAG and tool-use baseline.
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.
These two models have no overlapping published benchmarks in our dataset. Compare specs and pricing instead.