Is ERNIE-4.5-21B-A3B-Thinking or Qwen3.5-397B-A17B cheaper?

ERNIE-4.5-21B-A3B-Thinking is cheaper on output tokens ($0.28 vs $3.60 per 1M).

Which has the larger context window, ERNIE-4.5-21B-A3B-Thinking or Qwen3.5-397B-A17B?

Qwen3.5-397B-A17B has the larger context window (262K tokens).

ERNIE-4.5-21B-A3B-Thinking vs Qwen3.5-397B-A17B

ERNIE-4.5-21B-A3B-Thinking is cheaper on output tokens, while Qwen3.5-397B-A17B offers a larger context window. Choose ERNIE-4.5-21B-A3B-Thinking or Qwen3.5-397B-A17B based on the trade-off between cost, context, and the benchmarks that matter for your use case.

Spec	ERNIE-4.5-21B-A3B-Thinking	Qwen3.5-397B-A17B
Provider	NovitaAI	NovitaAI
Input / 1M tokens	$0.07	$0.60
Output / 1M tokens	$0.28	$3.60
Context window	131K	262K
Parameters	—	397B
Open weights	Yes	Yes
Released	Sep 2025	Feb 2026

ERNIE-4.5-21B-A3B-Thinking details →Qwen3.5-397B-A17B details →

FAQ

Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.