Grok 4.6 or Qwen3.8 Max 0902: which to choose?

What are we comparing?

This comparison pits the flagship models of two major developers against each other: Grok 4.6 and Qwen3.8 Max 0902, one from xAI, the other from Alibaba Cloud (Qwen). The first dates from August 12, 2026, the second from September 2, 2026.

All figures below come from public sources checked on October 22, 2026: models.dev and OpenRouter for prices and capabilities, Epoch AI for scores.

Performance

On our capability index, Grok 4.6 scores 86.1 out of 100 and Qwen3.8 Max 0902 88.4. Qwen3.8 Max 0902 leads by 2.3 points.

On the 3 benchmarks reported for both, Grok 4.6 wins 2 times and Qwen3.8 Max 0902 1:

• GPQA Diamond (PhD-level science questions): Grok 4.6 94%, Qwen3.8 Max 0902 92.3%
• OTIS Mock AIME (competition mathematics problems): Grok 4.6 99.2%, Qwen3.8 Max 0902 100%
• FrontierMath (very hard research-level mathematics): Grok 4.6 66%, Qwen3.8 Max 0902 65.6%

Pricing

Grok 4.6 costs $2 per million input tokens and $6 per million output tokens; Qwen3.8 Max 0902 costs $2 and $6.

For a typical agent task (read 20,000 tokens, write 2,000), expect $0.052 with Grok 4.6 versus $0.052 with Qwen3.8 Max 0902. The cost is almost the same.

Run the numbers with your own volumes in the agent cost calculator.

Features and limits

Qwen3.8 Max 0902 reads more in one go: 1,000,000 tokens of context (roughly 1,500 pages), versus 500,000 for Grok 4.6.

Reasoning mode: Grok 4.6 yes, Qwen3.8 Max 0902 yes. Tool calling: Grok 4.6 yes, Qwen3.8 Max 0902 yes. Image input: Grok 4.6 yes, Qwen3.8 Max 0902 yes.

Neither publishes its weights: both are used through an API.

In short

Our reading of the data:

• Choose Qwen3.8 Max 0902 for a higher capability index and a larger context window.

This verdict relies on public figures; it does not replace a trial on your own tasks.