Kimi K2.5

Kimi K2.5 sits within the kimi-k2 family and is positioned as a multimodal system that accepts text, image, and video inputs while producing text outputs. The combination of multimodal input with reasoning, tool calling, and structured output support points to a model intended for agent-style and complex analytical workloads rather than narrow single-task use. Open-weight availability further broadens its appeal for teams that want to run or fine-tune the model in their own environments rather than rely solely on hosted access.

Community discussion around Kimi K2.5 has drawn attention from power users of competing frontier assistants, with commentary noting the model's quality and the broader strategic context around Chinese-developed frontier systems. That reception, alongside the model's multimodal and tool-use capability set, suggests a practical fit for workflows that combine visual and textual information with structured, agentic execution. Practitioners evaluating the model will find the most utility in scenarios that benefit from reasoning over mixed media, programmatic tool orchestration, and structured response formats.

Basetenmoonshotai/Kimi-K2.5kimi-k2

Open provider page Provider docs

Quick Info

Provider: Baseten
Model key: moonshotai/Kimi-K2.5
Release date: Jan 30, 2026
Last updated: Feb 12, 2026
Knowledge cutoff: 2025-12
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.60
Output token cost: $3.00

Limits

Output tokens: 262,000 tokens
Context window: 262,000 tokens

Transparent token rates

Compare Kimi K2.5 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Kimi K2.5

Melious

CoverageBenchmark

Lorphic's July 2026 explainer walks through the Kimi K2 model family, explicitly distinguishing K2, K2.5, K2.6, and K2.7 as architecturally related but with distinct capabilities and licensing. The piece confirms every K2-line model shares a 1-trillion-parameter MoE design with 32B active parameters, 384 routed experts The article situates Kimi K2.5 within the broader release timeline as an intermediate family member, useful for practitioners deciding which K2 variant to target. It frames the MoE architecture as the reason Kimi models can be priced competitively against dense models of comparable scale, since inference cost tracks ro

Baseten

Official sourceBenchmark

OpenClaw + Kimi K2.5 on Baseten: frontier agent performance with open-source models. Outperforms Opus 4.5 and 8x cheaper. Install in 2min.

Ofox

Coverage

Moonshot AI released Kimi K2.5 in January 2026, an open-source model built on a Mixture-of-Experts architecture with 1 trillion total parameters and 32 billion activated per request. According to the Codecademy guide, it was trained on 15 trillion tokens that mixed visual and textual data from the start, letting vision The Codecademy article highlights Kimi K2.5's headline benchmark: 50.2% on Humanity's Last Exam at roughly 76% lower cost than Claude Opus 4.5, alongside claims of a 4.5x execution-time speedup from its parallel-agent approach. The Agent mode supports autonomous workflows with 200–300 tool calls. The piece positions K2