Hy3-preview vs MiniCPM-V 4.6 1.3B: Benchmarks, Pricing & Capabilities Compared
TL;DR — Hy3-preview wins for reasoning · MiniCPM-V 4.6 1.3B wins for long-context.
Hy3-preview Tencent
- Released
- 2026-04-23
- Context window
- 256K tokens
- Input price
- $0.00 / Mtok
- Output price
- $0.00 / Mtok
MiniCPM-V 4.6 1.3B OpenBMB
- Released
- 2026-05-11
- Context window
- 262K tokens
- Input price
- $0.00 / Mtok
- Output price
- $0.00 / Mtok
Benchmark comparison
| Benchmark | Hy3-preview | MiniCPM-V 4.6 1.3B |
|---|---|---|
| AA Intelligence Index | 41.9 ✓ | 12.7 |
| GPQA Diamond | 86.7% ✓ | 30.5% |
| HLE | 25.5% ✓ | 4.9% |
| IF-Bench | 63.1% ✓ | 26.7% |
| LiveCodeBench Reasoning | 54.7% ✓ | 6.3% |
| SciCode | 41.2% ✓ | 2.1% |
| TAU2-bench | 92.7% ✓ | 87.7% |
| TerminalBench-Hard | 34.1% ✓ | 0.0% |
Pricing comparison
| Metric | Hy3-preview | MiniCPM-V 4.6 1.3B |
|---|---|---|
| Input ($/Mtok) | $0.00 | $0.00 |
| Output ($/Mtok) | $0.00 | $0.00 |
| Cached input ($/Mtok) | — | — |
| Cost per 1M-token roundtrip (1M in + 1M out) | $0.00 | $0.00 |
Context window & modalities
| Attribute | Hy3-preview | MiniCPM-V 4.6 1.3B |
|---|---|---|
| Context window | 256K tokens | 262K tokens |
| Input modalities | text | text, image, video |
| Output modalities | text | text |
| Knowledge cutoff | — | — |
Verdict by use case
Coding
Insufficient data
Basis: SWE-bench
No shared coding benchmark.
Reasoning
→ Hy3-preview
Basis: GPQA Diamond
Hy3-preview 86.7% vs MiniCPM-V 4.6 1.3B 30.5% on GPQA Diamond.
Math
Insufficient data
Basis: MATH / AIME
No shared math benchmark.
Long context
→ MiniCPM-V 4.6 1.3B
Basis: Context window
Hy3-preview 256K tokens vs MiniCPM-V 4.6 1.3B 262K tokens.
Cost
Tie
Basis: Input $/Mtok
Hy3-preview $0/Mtok vs MiniCPM-V 4.6 1.3B $0/Mtok input.
Changelog & releases
Hy3-preview
Released 2026-04-23
MiniCPM-V 4.6 1.3B
Released 2026-05-11