Kimi VL A3B
Model overview
Kimi-VL-A3B-Instruct
Moonshot/Kimi's current public Kimi VL A3B efficient vision-language family.
multimodalvision-languagereasoningagentlong contextchatdocument understandinglocal inference
Representative version
Kimi-VL-A3B-Instruct
| Company | Moonshot AI / Kimi |
|---|---|
| Representative basis | Variant referenced by model provenance 2026-07-15 |
| Release | Not disclosed |
| Release type | Model release |
| Parameters | 16B / 3B active |
| Architecture | Not disclosed |
| Context | 128,000 |
| Input modalities | text / image / video |
| Output modalities | text |
| Tool use / function calling | Not disclosed |
| Structured output / reasoning | Not shown: Reasoning |
| Vision / audio / video | Confirmed: Vision, Video; Not shown: Audio |
| Open weights / License | Weights available / MIT License sourcemoonshot-kimi-kimi-vl-a3b-instruct-current · 2026-07-15 |
| Vendor API | Not disclosed API source pending verification |
| OpenRouter | Not separately listed on OpenRouter |
| Pricing | No reliable public source |
| Benchmark | No reliable public source |
| Public signals | 1 runtime / access record / 1 community discussion / 2 activity snapshots |
| Use cases | assistant chat / document understanding / local inference / multimodal apps |
| Claim evidence | Release / model specs / weightsmoonshotai/Kimi-VL-A3B-Instruct · Hugging FaceKimi-VL-A3B-Instruct |
OpenRouter
Pricing / context / availability
| OpenRouter | Not separately listed on OpenRouter |
|---|---|
| OpenRouter listing | Not separately listed in the current OpenRouter public list |
| Mapped version | Kimi-VL-A3B-Instruct moonshot-kimi-kimi-vl-a3b-instruct-current |
| Pricing | No reliable public source |
| Context | 128000 |
| Speed/Latency | No reliable public source |
| Snapshot time | 2026-07-15T20:11:41.551Z |
| Source | View source |
Public activity counters
Cumulative Hugging Face model-card downloads / likes / spaces
| Platform | Entry | Downloads | Likes | Spaces | Time | Source |
|---|---|---|---|---|---|---|
| huggingface | moonshotai/Kimi-VL-A3B-Instruct | 356,547 | 273 | 0 | 2026-07-15T20:11:41.551Z | View source |
| huggingface | moonshotai/Kimi-VL-A3B-Thinking | 145,891 | 449 | 24 | 2026-07-15T20:11:41.551Z | View source |
Operations
Runtime / access signals
| Time | Signal | Original text | Source |
|---|---|---|---|
| 06/06/2025, 01:07 PM UTC | <Think> is not generated. why? | I have tried 2 times on web demo. The output looks like "thoughts", but <Think is missing. According to the tech report, <Think or <think shou... | Source |
Vertical comparison
Current available versions
| Version | Release | Parameters | Context | Input | Output | Open weights / License | Vendor API | Source tier |
|---|---|---|---|---|---|---|---|---|
| Kimi-VL-A3B-Instruct | Not disclosed | 16B / 3B active | 128,000 | text / image / video | text | Weights available / MIT | Not disclosed | P0 |
| Kimi-VL-A3B-Thinking | Not disclosed | 16B / 3B active | 128,000 | text / image / video | text | Weights available / MIT | Not disclosed | P0 |
Evidence gaps
Public display boundary
Release date not disclosedOpenRouter not separately verifiedNo reliable benchmark sourceNo linkable real reviewSpeed/latency not disclosed