Kimi K3 vs Qwen3.8-Max

Qwen3.8-Max leads on the Visual Intelligence Index by 1.9 points, and Kimi K3 scores higher on 1 of the 7 benchmarks where the two differ.

Model

Compared with

At a glance

Measure	Kimi K3	Qwen3.8-Max
Visual Intelligence Index	71.9%	73.8%
Rank	#5 of 9	#4 of 9
Cost per taskMean over the benchmarks, at list price.	$0.114	$0.151
Latency per taskEach benchmark's median, averaged.	29.2s	66.0s
Tokens per task	31,619	68,932
Input / 1M tokens	$3.00	$2.00
Output / 1M tokens	$15.00	$6.00
Context window	1.05M	1M
Availability	Open weights	Proprietary

Each capability index is a weighted mean of the benchmarks that contribute to the capability.

Capability	Kimi K3	Qwen3.8-Max	Lead
Perception	71.5%	72.0%	−0.6
Spatial Reasoning	73.7%	74.3%	−0.6
Temporal Reasoning	72.4%	73.6%	−1.2
Action / Event Understanding	66.8%	71.5%	−4.7
Causal Reasoning	75.2%	78.2%	−3.0
Information Retrieval	70.1%	72.0%	−1.8

Benchmark	Kimi K3	Qwen3.8-Max	Lead
Video-MME v2	55.2%	60.2%	−5.1
LVBench	66.8%	58.9%	+7.9
Perception Test	82.4%	83.9%	−1.5
NExT-QA	86.3%	87.6%	−1.3
Q-Bench Video	67.5%	69.0%	−1.5
EgoSchema	78.6%	85.2%	−6.6
UCF101-AD	39.4%	56.5%	−17.1

4.0/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable1.0: You can run it yourself, and are not tied to one vendor.
Open weights — supported
Self-hostable — supported
Multimodal input0.5: It takes the footage directly, rather than frames you extracted.
Video ingestion — supported
Audio — not supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable0.5: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — not supported

3.5/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable0.0: You can run it yourself, and are not tied to one vendor.
Open weights — not supported
Self-hostable — not supported
Multimodal input0.5: It takes the footage directly, rather than frames you extracted.
Video ingestion — supported
Audio — not supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable1.0: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — supported