Compare AI models
| Specifications | ||
ParametersThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. | 1T | 0.8B/2B/4B/9B |
Context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider. | 256k | 256k |
| API pricingUSD per 1M tokens · lower wins | ||
Cheapest inputLowest input rate across every provider on OpenRouter, the lab included. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $0.49AtlasCloud | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $2.50AtlasCloud | — |
| $223AtlasCloud · int4 | — | |
| Benchmarks | ||
LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. | 85% | 65.6% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 87.6% | 81.7% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | 87.1% | 82.5% |
OSWorld-VerifiedAgentic computer use — Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. | 63.3% | 41.8% |
BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 52% | — |
SWE-Bench ProAgentic coding — Can the AI fix real bugs in real software? It's handed actual problems from open-source projects and has to write code that genuinely solves them. Higher is better. | 50.7% | — |
SWE-Bench VerifiedCoding — Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better. | 76.8% | — |
SWE-Bench MultilingualMultilingual coding — Like SWE-Bench, but the coding problems span many programming languages, not just one. Tests how broadly the AI can code. Higher is better. | 73% | — |
Next.js EvalsNext.js coding — Vercel's open eval of how well AI coding agents build and migrate real Next.js apps — measured as the share of tasks the agent completes successfully. Higher is better. | 18% | — |
Terminal-Bench 2.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? (Version 2.0 of the test.) Higher is better. | 50.8% | — |
IFEvalAgentic tasks — following verifiable instructions in generated responses | — | 91.5% |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | 60.6% | — |
Humanity's Last Exam · no toolsMultidisciplinary reasoning — Humanity's Last Exam — extremely hard expert questions across many subjects, written so you can't just look up the answer. “No tools” means the AI answers on its own. Higher is better. | 30.1% | — |
Humanity's Last Exam · with toolsMultidisciplinary reasoning — Humanity's Last Exam — extremely hard expert questions across many subjects. “With tools” means the AI is allowed to search the web or run code while answering. Higher is better. | 50.2% | — |
AIME 2025Reasoning — single-answer accuracy on the 2025 invitational mathematics exam without tools | 96.1% | — |
IFBenchInstruction following — Tests whether the AI can follow detailed instructions and satisfy multiple constraints at once. Higher is better. | — | 64.5% |
CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better. | — | 73% |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | — | 70.1% |
MMMUMultimodal — Tests the AI on understanding images and text together across many college subjects. Higher is better. | — | 78.4% |
| Overview | ||
| Company | Moonshot AI | Qwen |
| Release date | Jan 27 2026 | Mar 2 2026 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
Kimi K2.5vsClaude Fable 5.2Qwen3.5 SmallvsClaude Fable 5.2Kimi K2.5vsGPT-6.2 AstraQwen3.5 SmallvsGPT-6.2 AstraKimi K2.5vsGemini 4 NeonQwen3.5 SmallvsGemini 4 NeonKimi K2.5vsMuse EmberQwen3.5 SmallvsMuse EmberKimi K2.5vsGrok 4.7Qwen3.5 SmallvsGrok 4.7Kimi K2.5vsDeepSeek-V4.1-Flash-1017Qwen3.5 SmallvsDeepSeek-V4.1-Flash-1017Frequently asked questions
Kimi K2.5 leads Qwen3.5 Small on 4 of the 4 benchmarks they both report (LiveCodeBench, GPQA Diamond, MMLU-Pro, OSWorld-Verified). Kimi K2.5 shipped 34 days before Qwen3.5 Small, so benchmark comparisons should account for the intervening progress.
Kimi K2.5 has 1T parameters, while Qwen3.5 Small has 0.8B/2B/4B/9B. Context windows are 256k (Kimi K2.5) vs 256k (Qwen3.5 Small).
On LiveCodeBench, Kimi K2.5 leads at 85% vs Qwen3.5 Small at 65.6%. On GPQA Diamond, Kimi K2.5 leads at 87.6% vs Qwen3.5 Small at 81.7%. On MMLU-Pro, Kimi K2.5 leads at 87.1% vs Qwen3.5 Small at 82.5%. On OSWorld-Verified, Kimi K2.5 leads at 63.3% vs Qwen3.5 Small at 41.8%.