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. | — | 16B |
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. | 1M | 8k |
| Benchmarks | ||
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 91.9% | 35.2% |
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. | 48% | — |
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. | 43.3% | — |
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. | 76.2% | — |
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. | 50% | — |
LiveCodeBench ProCompetitive programming — Elo rating on competitive programming problems in LiveCodeBench Pro | 2439 | — |
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. | — | 14.3% |
HumanEvalFunction synthesis — 164 small Python problems: the AI is given a function's description and has to write the working function. This was the coding benchmark of the GPT-3.5 and GPT-4 era, before the field moved to fixing real bugs in real repositories. Higher is better. | — | 48.1% |
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. | 54.2% | — |
APEX-AgentsExpert agentic work — expert-level agentic work tasks (AI Productivity Index) | 18.4% | — |
MCP Atlas · original harnessMulti-step tool use — pass rate under the original tool-use harness, before Scale changed the judge and retry handling in April 2026 | 54.1% | — |
τ²-bench · retailAgentic tasks — task success in the retail domain of the second-generation customer-service evaluation | 85.3% | — |
τ²-bench · telecomAgentic tasks — task success in the telecom domain of the second-generation customer-service evaluation | 98% | — |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | 59.2% | — |
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. | 37.5% | — |
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. | 45.8% | — |
ARC-AGI-2Abstract reasoning — Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better. | 31.1% | — |
MATHReasoning — accuracy on the original competition mathematics test set | — | 45.3% |
WinoGrandeReasoning — resolving ambiguous pronouns in context | — | 74.6% |
AIME 2025Reasoning — single-answer accuracy on the 2025 invitational mathematics exam without tools | 95% | — |
SimpleQA VerifiedKnowledge — accuracy on the vetted factual questions in SimpleQA Verified | 72.1% | — |
SimpleQAKnowledge — accuracy on short, fact-seeking questions in the original SimpleQA set | — | 5.6% |
MMMLUKnowledge — average accuracy on multilingual translations of MMLU | 91.8% | — |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | — | 42.4% |
MMLUGeneral knowledge — A 57-subject multiple-choice exam — history, law, medicine, maths — that was the standard measure of how much a model knows from 2020 until roughly 2024, when frontier scores crowded into the high 80s and labs moved on to harder tests. The scores here were published years apart under different testing setups, so read them as a historical record rather than a like-for-like ranking. Higher is better. | — | 70% |
GSM8KGrade-school math — Grade-school maths word problems that take a few steps of arithmetic to work through. It separated the models of 2022 and 2023 sharply, then saturated. One caveat on the historical numbers: OpenAI included part of the GSM8K training set in GPT-4's pre-training mix, so GPT-4's score is not a clean few-shot result. Higher is better. | — | 77.4% |
CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better. | 81.4% | — |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | 81% | — |
MRCR v2 (8-needle) · 128k averageLong context — Tests whether the AI can find specific details buried inside a very long document (around 128k tokens — roughly a long book). Higher is better. | 77% | — |
MRCR v2 (8-needle) · 1M pointwiseLong context — Tests whether the AI can find specific details buried inside an enormous document (around 1 million tokens — many books). Higher is better. | 26.3% | — |
| Overview | ||
| Company | Moonshot AI | |
| Release date | Nov 18 2025 | Feb 22 2025 |
| Access | Closed | Open Weight |
| Model details | View model | View model |
Other comparisons
Gemini 3.0 ProvsClaude Fable 5.2MoonlightvsClaude Fable 5.2Gemini 3.0 ProvsGPT-6.2 AstraMoonlightvsGPT-6.2 AstraGemini 3.0 ProvsMuse EmberMoonlightvsMuse EmberGemini 3.0 ProvsGrok 4.7MoonlightvsGrok 4.7Gemini 3.0 ProvsDeepSeek-V4.1-Flash-1017MoonlightvsDeepSeek-V4.1-Flash-1017Gemini 3.0 ProvsMistral Large 4MoonlightvsMistral Large 4Frequently asked questions
Gemini 3.0 Pro leads Moonlight on 1 of the 1 benchmark they both report (GPQA Diamond). Moonlight shipped 269 days before Gemini 3.0 Pro, so benchmark comparisons should account for the intervening progress.
Context windows are 1M (Gemini 3.0 Pro) vs 8k (Moonlight). Gemini 3.0 Pro is closed, while Moonlight is open weight.
On GPQA Diamond, Gemini 3.0 Pro leads at 91.9% vs Moonlight at 35.2%.