OpenGPTEvidence-based model decisionsStarting candidate 1
StableQwen 3.7 Max
qwen3.7-max-2026-06-08Alibaba QwenNo comparable public result is currently mapped to this exact version.
06 · TASK RANKING
Chinese & multilingual work
Chinese-first writing, translation, localization, and mixed-language tasks.
- Official version
- 3
- Public evidence
- 2
- Data cutoff:
- Why these candidates
- Start with providers that publish Chinese-focused or multilingual model lines, then evaluate terminology and regional language with your own samples.
- What remains uncertain
- The public overall ranking is not a Chinese-language benchmark; language quality must be tested directly.
- Public evidence
- User preference
- Data cutoff
Starting candidate
3 / 3Three official versions to test first
Public data narrows the field. It does not replace a controlled test on your own work.
Starting candidate 2
StableKimi K3
kimi-k3Moonshot AIUser preference · #10
Starting candidate 3
PreviewDeepSeek V4 Pro
deepseek-v4-proDeepSeekUser preference · #46
Public evidenceThe models use different public evidence scales. Missing evidence is not treated as zero, and OpenGPT does not combine unlike sources into one score.⌄
A practical starting point, not a universal winnerSelect two or three candidates for a side-by-side public data comparison.
The models use different public evidence scales. Missing evidence is not treated as zero, and OpenGPT does not combine unlike sources into one score.
The public overall ranking is not a Chinese-language benchmark; language quality must be tested directly.
The evaluation opens with this task template and the first two selected models. You can edit every task before collecting answers.