Pyyan / Compare / Hy-MT2-30B-A3B vs GPT-6 Astra
Hy-MT2-30B-A3B vs GPT-6 Astra
2 of 5Language Models · verified 5 Sept 2026
Hy-MT2-30B-A3BTencentcurrent
GPT-6 AstraOpenAIcurrent| Specification | Hy-MT2-30B-A3B | GPT-6 Astra |
|---|---|---|
| Summary | A translation specialist across 33 language pairs, priced like nothing else. | Saturates three of the benchmarks it reports, and ships disabled for enterprise. |
| Context | 8K | 1.05M |
| Input ($/Mtok) | $0.074 | $10 |
| Output ($/Mtok) | $0.295 | $50 |
| Licence | Proprietary | Proprietary |
| Max output | 8K | 128K |
| Parameters | Not published | Not published |
| Released | 20 Aug 2026 | 3 Sep 2026 |
| Model ID | hy-mt2-30b-a3b | gpt-6-astra |
| Category | Language Models | Language Models |
| Official | — | — |
Highlighted rows are where these differ.
Hy-MT2-30B-A3B
- Tencent's translation flagship, covering 33 language pairs
- Seven cents per million input tokens, which is roughly a hundredth of a frontier model
- Short 8K context, because translation does not need more
Best for translation at volume.
GPT-6 Astra
- 97.6% on FrontierMath Tier 4, 99.9% on ARC-AGI-3 and 100% on ExploitBench
- Above 272K input tokens the whole request reprices: input and cache at 2x, output at 1.5x
- Rolls out to OpenAI's Daybreak cyber programme first, and arrives switched off for enterprise
Best for hard reasoning and computer use.