OnlySOTA

Qwen 3.8 Flash Next vs GPT 5.6 Terra

Both are SOTA right now. Each row below is one board’s call on the two, and the marked figure is the one that board places higher. No row is added to another, so there is no overall winner: the boards measure different things, and where they split, that is the finding.

Board Qwen 3.8 Flash Next GPT 5.6 Terra
Artificial Analysis LLM Leaderboard Artificial Analysis Intelligence Index 39.8 No. 24 of 372 42.1 No. 19 of 372
Arena — Agent Overall -0.007 No. 35 of 49 0.004 No. 32 of 49
Arena — Code Overall 1637 No. 16 of 109 1519 No. 46 of 109
Arena — Text Overall (style control) — 1465 No. 51 of 306
Arena — Vision Overall (style control) — 1268 No. 36 of 128

Each rank is the board’s own: the rank it publishes, or its place in its own order where it publishes none. The bar is how much of that board the model is ahead of. Scores from different boards cannot be compared.

At a glance

Developer Alibaba OpenAI
Weights open closed
Released 2026-08-262026-07-09
Price Blended Price, USD per 1M tokens (3:1 input:output) $0.23$4.50
Value: where it sits on the kill line SOTA SOTA

Qwen 3.8 Flash Next →GPT 5.6 Terra → How the kill line is drawn →

Qwen 3.8 Flash Next against the other SOTA models

GPT 5.6 Terra against the other SOTA models