Head-to-head comparison
KAT-Coder-Pro V2.5 (Kuaishou) and Qwen3 VL 235B A22B Thinking (Qwen) compared on the ModelCap Index, API price, context window, provider availability, weight access and every public benchmark board they share. Figures are the same ones shown on the live rankings; nothing here is a hidden score.
Snapshot as of 3 September 2026
As of 3 September 2026, Qwen3 VL 235B A22B Thinking holds the stronger ModelCap Index position (#59 vs #61); KAT-Coder-Pro V2.5 is 1.4× cheaper per output token ($2.96/1M vs $4.00/1M); KAT-Coder-Pro V2.5 offers the longer context window (262K vs 131K tokens); and Qwen3 VL 235B A22B Thinking ships open weights.
Evidence: start rank inherited from the measured predecessor in its own product family, less a fixed succession penalty, until independent measurements replace it · 2 public benchmark observations across 2 boards. Prices are the lowest listed API offer per million tokens observed on the OpenRouter catalogue.
Scores are the sources' own published figures for each model's best evaluated configuration; ModelCap never re-runs a benchmark.
Want a different pairing? Open the interactive comparison tool to swap either model for any current ranked language model.
Qwen3 VL 235B A22B Thinking ranks higher on the ModelCap Index as of 3 September 2026: #59 against #61. That is a capability ranking built from public benchmark evidence with published uncertainty; whether it is "better" for you also depends on price, context and where you can run it.
KAT-Coder-Pro V2.5 is cheaper on output tokens: $2.96/1M against $4.00/1M. Input tokens are $0.74/1M for KAT-Coder-Pro V2.5 and $0.40/1M for Qwen3 VL 235B A22B Thinking. Prices are the lowest listed API offer ModelCap observed, in USD per million tokens.
KAT-Coder-Pro V2.5 has the larger context window: 262K tokens against 131K.
The two models do not share a coding benchmark board on ModelCap yet, so no head-to-head coding score is published; the ModelCap Index position is the closest overall signal.
KAT-Coder-Pro V2.5: API only. Qwen3 VL 235B A22B Thinking: Open weights. Open weights mean the checkpoint can be downloaded and self-hosted under its licence; API-only models are available solely through hosted endpoints.
ModelCap currently lists 1 API provider for KAT-Coder-Pro V2.5 and 2 for Qwen3 VL 235B A22B Thinking, from the OpenRouter catalogue snapshot the site serves; each model page lists the providers and their prices.
Every figure comes from the sealed ModelCap dataset published 3 September 2026; the page re-renders within a minute of each data refresh, and the ModelCap Index positions are the same ones shown on the live rankings.