Kimi K2.5
Moonshot AIhuggingface.co/moonshotai/Kimi-K2.5
Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm. Built on Kimi K2 with continued pretraining over approximately 15T mixed...
Updated 28 min ago · first seen 11 Sept 2026
model_01M294WW5TTBRH9A6F6RZ9EH1W
- Parameters
- 1.03T
- T2 · 3 h ago
- Context
- 256K tokens
- T2 · 1 h ago
- Released
- 1 Jan 2026
- T2 · 28 min ago
- License
- other
- T2 · 2 h ago
Specification
- Release date
- 1 Jan 2026
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 28 min agomedium
- Status
- deprecated
Source:Artificial AnalysisT2observed 1 h agomedium
- Openness
- open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- License
- other
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Architecture
- KimiK25ForConditionalGeneration
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Parameters
- 1.03T
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- Context window
- 256K tokens
Source:Artificial AnalysisT2observed 1 h agomedium
- Max output
- 235.9K tokens
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Modalities
- image, text
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Input modalities
- image, text
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Output modalities
- text
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- File size
- 595.2 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Hugging Face repo
- moonshotai/Kimi-K2.5
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Pipeline tag
- image-text-to-text
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- Model card
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- Downloads
- 382,523
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 28 min agomedium
- Likes
- 2,869
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 28 min agomedium
- Gated
- No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Hf inference providers
- deepinfra, featherless-ai, novita
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Last modified
- 2026-04-30T03:56:40+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- Library name
- transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- License name
- modified-mit
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Downloads all time
- 15,726,724
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 28 min agomedium
- Model type
- kimi_k25
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Openrouter id
- moonshotai/kimi-k2.5
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Supported parameters
- frequency_penalty, include_reasoning, logit_bias, logprobs, max_tokens, min_p, presence_penalty, reasoning, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_logprobs, top_p
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
- Tags
- transformers, safetensors, kimi_k25, feature-extraction, compressed-tensors, image-text-to-text, custom_code
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Weights dtype
- BF16, F32, I32
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
33
Source tiers
T233
Freshest observation
28 min ago
Conflicts
None
Modalities
- Modalities
- imagetext
- Input
- imagetext
- Output
- text
Capabilities
Tool calling
Yes
OpenRouter public model & pricing listing · T2
Structured output
Yes
OpenRouter public model & pricing listing · T2
Reasoning
Yes
OpenRouter public model & pricing listing · T2
Vision
Yes
OpenRouter public model & pricing listing · T2
Audio
Unavailable
Fine-tuning available
Unavailable
- Context window
- 256K tokens
Source:Artificial AnalysisT2observed 1 h agomedium
- Max output
- 235.9K tokens
Source:OpenRouter public model & pricing listingT2observed 3 h agomedium
| Benchmark | Score | Metric | Config | Evaluated | Source |
|---|---|---|---|---|---|
| τ²-bench | 95.91% | pass^1 | aa_slug=kimi-k2-5 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Artificial Analysis Intelligence Index | 23.46 | index | aa_slug=kimi-k2-5 · version=4.3 · estimated=true | — | artificialanalysis.aiT2 |
| GPQA | 87.88% | accuracy | aa_slug=kimi-k2-5 · variant=GPQA Diamond · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Humanity's Last Exam | 30.72% | accuracy | aa_slug=kimi-k2-5 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| IFBench | 70.2% | accuracy | aa_slug=kimi-k2-5 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| MMMU-Pro | 75.38% | accuracy | aa_slug=kimi-k2-5 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| SWE-bench Multilingual | 67.3% | resolved | date=2026-02-13 · board=Multilingual · system=mini-SWE-agent · model_tag=kimi-k2.5 | 13 Feb 2026 | swebench.comT2 |
| SWE-bench Verified | 70.8% | resolved | date=2026-02-17 · board=Verified · system=mini-SWE-agent · model_tag=kimi-k2.5 | 17 Feb 2026 | swebench.comT2 |
| Terminal-Bench | 34.85% | accuracy | aa_slug=kimi-k2-5 · variant=hard · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 45.69% | accuracy | aa_slug=kimi-k2-5 · variant=v2.1 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
Scores are reported as published, with their evaluation configuration (harness, prompting, judge). Results with different configs are not directly comparable — see methodology.
Current prices
| Provider | Input / 1M | Output / 1M | Cached in | Batch in / out | Context | Observed | Source |
|---|---|---|---|---|---|---|---|
| Moonshot AI Platformmoonshotai/kimi-k2.5 | $0.45 | $2.25 | $0.07 | — | 262.1K | 1 h ago | openrouter.aiT2 |
USD per 1M tokens as published by each provider (USD). Rows are append-only: every change is kept in the history below.
Available through 1
Price history
Input price · USD / 1M tokens
Output price · USD / 1M tokens
- Moonshot AI Platform
Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.
| Hardware | Quantization | Memory | Est. need | Fits |
|---|---|---|---|---|
| NVIDIA DGX B200 | 4bit | 1,440 GB | 591.0 GB | Yes |
| Apple M3 Ultra | 4bit | — | 591.0 GB | No |
| Mac Studio (Apple M5 Ultra) | 4bit | — | 591.0 GB | No |
| AMD Instinct MI325X | 4bit | 256 GB | 591.0 GB | No |
| AMD Instinct MI300X | 4bit | 192 GB | 591.0 GB | No |
| Apple M2 Ultra | 4bit | — | 591.0 GB | No |
| NVIDIA B200 | 4bit | 180 GB | 591.0 GB | No |
| NVIDIA H200 | 4bit | 141 GB | 591.0 GB | No |
| NVIDIA H200 NVL | 4bit | 141 GB | 591.0 GB | No |
| Apple M1 Ultra | 4bit | — | 591.0 GB | No |
| Apple M3 Max | 4bit | — | 591.0 GB | No |
| Apple M4 Max | 4bit | — | 591.0 GB | No |
| Mac Studio (Apple M5 Max) | 4bit | — | 591.0 GB | No |
| MacBook Pro (Apple M5 Max) | 4bit | — | 591.0 GB | No |
| NVIDIA DGX Spark | 4bit | 128 GB | 591.0 GB | No |
| Apple M2 Max | 4bit | — | 591.0 GB | No |
| NVIDIA H100 NVL | 4bit | 94 GB | 591.0 GB | No |
| NVIDIA A100 80GB | 4bit | 80 GB | 591.0 GB | No |
| NVIDIA H100 SXM | 4bit | 80 GB | 591.0 GB | No |
| Apple M1 Max | 4bit | — | 591.0 GB | No |
| Apple M4 Pro | 4bit | — | 591.0 GB | No |
| Mac mini (Apple M5 Pro) | 4bit | — | 591.0 GB | No |
| MacBook Pro (Apple M5 Pro) | 4bit | — | 591.0 GB | No |
| Apple M3 Pro | 4bit | — | 591.0 GB | No |
| Apple M1 Pro | 4bit | — | 591.0 GB | No |
| Apple M2 Pro | 4bit | — | 591.0 GB | No |
| Apple M4 | 4bit | — | 591.0 GB | No |
| iMac (Apple M4) | 4bit | — | 591.0 GB | No |
| Mac mini (Apple M6) | 4bit | — | 591.0 GB | No |
| MacBook Air (Apple M5) | 4bit | — | 591.0 GB | No |
| MacBook Pro (Apple M5) | 4bit | — | 591.0 GB | No |
| NVIDIA GeForce RTX 5090 | 4bit | 32 GB | 591.0 GB | No |
| Apple M2 | 4bit | — | 591.0 GB | No |
| Apple M3 | 4bit | — | 591.0 GB | No |
| NVIDIA GeForce RTX 3090 | 4bit | 24 GB | 591.0 GB | No |
| NVIDIA GeForce RTX 4090 | 4bit | 24 GB | 591.0 GB | No |
| Apple M1 | 4bit | — | 591.0 GB | No |
Ancestors 0
None recorded.
Descendants 0
None recorded.
Papers 1
- arXiv:2602.02276Active35
Repositories 0
No repositories linked yet.
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history · Release date
Release daterelease_date4
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 1 Jan 2026 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
| 27 Jan 2026 | → | superseded | OpenRouter public model & pricing listingT2 | medium | deterministic |
| 1 Jan 2026 | → | superseded | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
| 27 Jan 2026 | → | superseded | OpenRouter public model & pricing listingT2 | medium | deterministic |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
Kimi K2.5 scores 70.8% on SWE-bench Verified
swebench_leaderboardKimi K2.5 scores 67.3% on SWE-bench Multilingual
swebench_leaderboardKimi K2.5 scores 34.85% on Terminal-Bench
artificial_analysisKimi K2.5 scores 45.69% on Terminal-Bench
artificial_analysisKimi K2.5 scores 30.72% on Humanity's Last Exam
artificial_analysisKimi K2.5 scores 23.46 on Artificial Analysis Intelligence Index
artificial_analysisKimi K2.5: context length changed from 262144 to 256000
Context window262.1K tokens→256K tokensartificial_analysisKimi K2.5: release date changed from 2026-01-01 to 2026-01-27
Release date1 Jan 2026→27 Jan 2026openrouterKimi K2.5: release date changed from 2026-01-27 to 2026-01-01
Release date27 Jan 2026→1 Jan 2026huggingfaceMoonshot AI Platform lists Kimi K2.5 at $0.45 in / $2.25 out per 1M tokens
openrouter
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/moonshotai/Kimi-K2.5 | model_page | T2· Quality secondary | 28 min ago | 3 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/mlx-community/Kimi-K2.5 | model_page | T2· Quality secondary | 28 min ago | 2 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/amd/Kimi-K2.5-MXFP4 | model_page | T2· Quality secondary | 50 min ago | 3 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/models?author=moonshotai&p=0&sort=downloads | listing | T2· Quality secondary | 54 min ago | 3 |
| SWE-bench leaderboards | swebench.com/ | leaderboard | T2· Quality secondary | 59 min ago | 1 |
| Artificial Analysis | artificialanalysis.ai/leaderboards/models | leaderboard | T2· Quality secondary | 1 h ago | 1 |
| OpenRouter public model & pricing listing | openrouter.ai/api/v1/models | catalogue | T2· Quality secondary | 1 h ago | 2 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/moonshotai/Kimi-K2.5/raw/main/README.md | model_card | T2· Quality secondary | 1 h ago | 1 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.