vs
8:1
overall capability wins (0 ties across 9 scored). Hunyuan 3 Preview leads overall.
- Reasoning5:1
- Coding2:0
- Pricing3:0
- Specs1:0
Verdict
Hunyuan 3 Preview leads coding (widest gap: +189.4 on WebDev Elo); Hunyuan 3 Preview offers ~2.0x larger context (262K vs 128K); Hunyuan 3 Preview is ~3.8x cheaper on a blended token basis than Mercury 2.
- Hunyuan 3 Preview leads coding (widest gap: +189.4 on WebDev Elo)
- Hunyuan 3 Preview offers ~2.0x larger context (262K vs 128K)
- Hunyuan 3 Preview is ~3.8x cheaper on a blended token basis than Mercury 2
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Hunyuan 3 Preview
- $0.12
- Mercury 2
- $0.44
Hunyuan 3 Preview is about 3.8x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Tencent
Inception Labs
Family
Other
Other
License
Proprietary
Proprietary
Open weights
No
No
Release
May 20, 2026
Sep 1, 2026
Knowledge cutoff
-
-
API / provider
Tencent
Inception Labs
Modalities
text, image → text
text → text
Specs
Context window
262K
128K
Max output
16K
-
Parameters
—
—
Pricing
Input $/1M
$0.06
$0.25
Output $/1M
$0.21
$0.75
Blended $/1M (3∶1)
$0.10
$0.38
Speed
tok/s
100
-
TTFT (s)
0.3
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
86.7
77.0
Humanity's Last Exam
27.8
17.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
27.9
21.2
AA-LCR v1.1
64.7
43.7
CritPt
4.6
0.8
MMMU-Pro
—
—
IFBench
63.1
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
41.2
37.7
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,356
1,167
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,413
—
| Metric | Hunyuan 3 Preview | Mercury 2 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Tencent | Inception Labs | — |
| Family | Other | Other | — |
| License | Proprietary | Proprietary | — |
| Open weights | No | No | — |
| Release | May 20, 2026 | Sep 1, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Tencent | Inception Labs | — |
| Modalities | text, image → text | text → text | — |
| Specs | |||
| Context window | 262K | 128K | A +134K |
| Max output | 16K | - | — |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $0.06 | $0.25 | A +$0.19 |
| Output $/1M | $0.21 | $0.75 | A +$0.54 |
| Blended $/1M (3∶1) | $0.10 | $0.38 | A +$0.28 |
| Speed | |||
| tok/s | 100 | - | — |
| TTFT (s) | 0.3 | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 86.7 | 77.0 | A +9.7 pts |
| Humanity's Last Exam | 27.8 | 17.1 | A +10.7 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 27.9 | 21.2 | A +6.7 pts |
| AA-LCR v1.1 | 64.7 | 43.7 | A +21.0 pts |
| CritPt | 4.6 | 0.8 | A +3.8 pts |
| MMMU-Pro | — | — | — |
| IFBench | 63.1 | 69.8 | B +6.7 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 | 41.2 | 37.7 | A +3.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,356 | 1,167 | A +189 Elo |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,413 | — | — |
Hunyuan 3 Preview
- Context
- 262K / 16K out
- Parameters
- —
- Price
- $0.06 / $0.21
- Speed
- 100 tok/s · 0.3s TTFT
- Modalities
- text, image → text
- License
- Proprietary
Next-gen Hunyuan preview SKU with long multimodal context and aggressive Flash-like pricing.
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.