Qwen3.8-27B vs Kimi K3

Kimi K3 leads on the Visual Intelligence Index by 3.4 points, and Qwen3.8-27B scores higher on 3 of the 7 benchmarks where the two differ.

Model

Compared with

At a glance

Measure	Qwen3.8-27B	Kimi K3
Visual Intelligence Index	68.5%	71.9%
Rank	#7 of 10	#6 of 10
Cost per taskMean over the benchmarks, at list price.	—	$0.114
Latency per taskEach benchmark's median, averaged.	32.0s	29.2s
Tokens per task	25,009	31,619
Input / 1M tokens	—	$3.00
Output / 1M tokens	—	$15.00
Context window	262K	1.05M
Availability	Open weights	Open weights

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

Capability	Qwen3.8-27B	Kimi K3	Lead
Perception	67.8%	71.5%	−3.6
Spatial Reasoning	68.6%	73.7%	−5.1
Temporal Reasoning	68.4%	72.4%	−4.0
Action / Event Understanding	65.2%	66.8%	−1.7
Causal Reasoning	72.3%	75.2%	−2.9
Information Retrieval	67.0%	70.1%	−3.2

Benchmark	Qwen3.8-27B	Kimi K3	Lead
Video-MME v2	50.5%	55.2%	−4.6
LVBench	64.6%	66.8%	−2.2
Perception Test	72.3%	82.4%	−10.1
NExT-QA	82.2%	86.3%	−4.1
Q-Bench Video	67.9%	67.5%	+0.3
EgoSchema	79.1%	78.6%	+0.5
UCF101-AD	45.5%	39.4%	+6.1

4.5/ 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 input1.0: It takes the footage directly, rather than frames you extracted.
Video ingestion — supported
Audio — 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

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