Kimi K2.5
kimi-k2-5kimi-k2.5- π§ Trillion-parameter Mixture-of-Experts, 32B active per token.
- π Native multimodal: continual pretraining on ~15T mixed vision-text tokens.
- π 256K-token context window for long documents and codebases.
- ποΈ Adds vision and cross-modal reasoning over text-only Kimi K2.
- π§ Instant and thinking modes for speed or deeper reasoning.
- β‘ Native INT4 quantization for efficient inference.
- π― Provider-reported HLE with tools: 51.8 text, 39.8 image.
- π Open-weight with OpenAI/Anthropic-compatible API.
Moonshot is an AI research lab known for developing the Kimi family of large language models. The organization has gained recognition for building capable reasoning-oriented models, with the Kimi line representing its flagship series of text generation systems.
Explore 8 more models by Moonshot βKimi K2.5, released January 2026 by Beijing-based Moonshot AI, is the multimodal evolution of the Kimi K2 line. It keeps the family's trillion-parameter Mixture-of-Experts design β with roughly 32 billion active parameters per token and a 256K context window β while adding a vision pathway. Per Moonshot's model card, K2.5 was built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base, so vision and language develop together rather than as bolted-on features.
The clearest gain over the text-only predecessor Kimi K2 is native multimodality: K2.5 understands images, supports cross-modal reasoning, and grounds agentic tool use in visual inputs. It also formalizes dual operation β an instant mode for speed and a thinking mode for deeper reasoning. On Moonshot's reported Humanity's Last Exam, K2.5 scores 31.5 text and 21.3 image without tools, rising to 51.8 text and 39.8 image with tools.
K2.5 became the architectural template for its successors. Kimi K2.6 reuses the same trillion-parameter, 32B-active topology, changing the post-training recipe, while Kimi K2.7 Code specializes the same MoE backbone for long-horizon software engineering. All ship open-weight with native INT4 quantization and an OpenAI/Anthropic-compatible API.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usageΒ Β·Β Kimi K2.5Β Β·Β Organic providers
Top providersΒ Β·Β Kimi K2.5Β Β·Β Organic
- Open Forge61.7%
- Phala30.2%
- β² Apex Ant4.3%
- Venice.ai Proxy3.6%
- klinn0.0%
- Super Seeder0.0%
- Others0.1%
| Provider | Health | Repβ | Sales | Tokens | Routing | Input $/M | Cached $/M | Output $/M | Save | 1st token | tok/s |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Relay Works 0x3040β¦7139 | β | 58.03 | $0.01 | 10.50k | gated | $0.325 | $0.0975 | $1.625 | β35% | β | β |
| HallPass 0xcac8β¦71b4 | β | 20.00 | β | β | gated | $0.40 | $0.12 | $2.00 | β20% | β | β |
| klinn 0x0ee1β¦6c41 | β | 12.34 | $0.03 | 794.01k | gated | $0.0156 | $0.0132 | $0.0779 | β97% | β | β |
| surplus-hive 0x6145β¦5c58 | β | 0.00 | β | β | gated | $1.0321 | $1.0321 | $5.1607 | β | β | β |
| Open Bird 0xc0f1β¦8183 | β | 0.00 | $0.00 | 944.00 | gated | $0.42 | $0.07 | $2.10 | β16% | β | β |
| Fire Ant π₯π 0xbe05β¦bc5d | β | 0.00 | $0.00 | 10.06k | gated | $0.41 | $0.198 | $3.15 | β18% | β | β |
| Leftermute 0x388bβ¦5389 | β | 0.00 | β | β | gated | $0.2828 | $0.2828not advertised | $1.7675 | β43% | β | β |
| Joi Gateway 0x8a90β¦7007 | β | 0.00 | β | β | gated | $0.2196 | $0.2196 | $1.098 | β56% | β | β |
| antseed-neon-puma-944e 0x6650β¦944e | β | 0.00 | β | β | gated | $0.191 | $0.22 | $1.1935 | β62% | β | β |
| Apex TEE Test 0xe672β¦7955 | β | 0.00 | β | β | gated | $0.0815 | $0.0315 | $0.3703 | β84% | β | β |
| Meridian AI 0x8c8cβ¦06f5 | β | 0.00 | $0.00 | 98.00 | gated | $0.0798 | $0.0798not advertised | $0.4309 | β84% | β | β |
| surplus-hive 0xe177β¦e91c | β | 0.00 | β | β | gated | $0.0317 | $0.0317 | $0.1583 | β94% | β | β |
| NovaRoute AI 0xc50dβ¦ed7b | β | 0.00 | $0.00 | 37.00 | gated | $0.0229 | $0.0508 | $0.1039 | β95% | β | β |
| Best 0x472fβ¦69fd | healthy | 0.00 | $0.02 | 354.12k | gated | $0.0156 | $0.0132 | $0.0779 | β97% | 0.2 s | 210 |
| β² Apex Ant 0x73b4β¦e736 | failingmodel gone Β· 13 h Β· stale | 88.20 | $2.65 | 157.46M | #1 | $0.02 | $0.0043 | $0.09 | β96% | β | β |
"Best price" and the provider table are live Antseed catalog data (advertised $/1M tokens β or $ per generated image for unit-billed image models β not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what Antseed Desktop ships with): a trust β₯ 60 gate on the effective reputation, then cheapest-first among routable providers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.