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. | 200k | 8k |
| API pricingUSD per 1M tokens · lower wins | ||
Input priceWhat you pay for everything you send the model — your question, plus any documents or earlier conversation you include with it. | $3.00 | — |
Output priceWhat you pay for the text the model writes back. It is normally the dearer half: producing an answer costs more than reading one. | $15.00 | — |
Cached input priceA reduced rate for text you send over and over. If every request starts with the same instructions or the same document, the provider keeps a copy ready and charges less to read it again. | $0.30 | — |
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. | $3.00Amazon Bedrock | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $15.00Amazon Bedrock | — |
| $1,350Anthropic list price | — | |
| Benchmarks | ||
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 83.4% | 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. | 79% | — |
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. | 77.2% | — |
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. | 36% | — |
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. | 51% | — |
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 | 43.8% | — |
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. | 17.7% | — |
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. | 33.6% | — |
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. | 13.6% | — |
MATHReasoning — accuracy on the original competition mathematics test set | — | 45.3% |
WinoGrandeReasoning — resolving ambiguous pronouns in context | — | 74.6% |
SimpleQAKnowledge — accuracy on short, fact-seeking questions in the original SimpleQA set | — | 5.6% |
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% |
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. | 61.4% | — |
GDPval-AAKnowledge work — Measures how well the AI does economically valuable knowledge work, judged against human experts. Shown as a rating (like a chess Elo) — higher is better. | 1276 | — |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | 63.4% | — |
MMMUMultimodal — Tests the AI on understanding images and text together across many college subjects. Higher is better. | 77.8% | — |
| Overview | ||
| Company | Anthropic | Moonshot AI |
| Release date | Sep 29 2025 | Feb 22 2025 |
| Access | Closed | Open Weight |
| Model details | View model | View model |
Other comparisons
Claude Sonnet 4.5vsGPT-6.2 AstraMoonlightvsGPT-6.2 AstraClaude Sonnet 4.5vsGemini 4 HydrogenMoonlightvsGemini 4 HydrogenClaude Sonnet 4.5vsMuse EmberMoonlightvsMuse EmberClaude Sonnet 4.5vsGrok 4.7MoonlightvsGrok 4.7Claude Sonnet 4.5vsDeepSeek-V4.1-Flash-1017MoonlightvsDeepSeek-V4.1-Flash-1017Claude Sonnet 4.5vsMistral Large 4MoonlightvsMistral Large 4Frequently asked questions
Claude Sonnet 4.5 leads Moonlight on 1 of the 1 benchmark they both report (GPQA Diamond). Only Claude Sonnet 4.5 has a verified first-party price for an available API: $3.00 per million input tokens and $15.00 per million output tokens. No currently available pay-as-you-go API rate is tracked for Moonlight. Moonlight shipped 219 days before Claude Sonnet 4.5, so benchmark comparisons should account for the intervening progress.
Context windows are 200k (Claude Sonnet 4.5) vs 8k (Moonlight). Claude Sonnet 4.5 is closed, while Moonlight is open weight.
On GPQA Diamond, Claude Sonnet 4.5 leads at 83.4% vs Moonlight at 35.2%.