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. | — | 1.5B/7B/72B |
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. | 400k | 4k |
| 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. | $0.20 | — |
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. | $1.25 | — |
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. | $0.02 | — |
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. | $0.10OpenAI | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $0.625OpenAI | — |
| $103OpenAI list price | — | |
| These models have no shared benchmark scores. | ||
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. | 14% | — |
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. | 52.4% | — |
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. | 46.3% | — |
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 | 56.1% | — |
ToolathlonGeneral tool use — Tests how well the AI uses everyday real-world tools and apps to get things done. Higher is better. | 35.5% | — |
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. | 24.3% | — |
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. | 37.7% | — |
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. | 5.7% | — |
MATHReasoning — accuracy on the original competition mathematics test set | — | 85.9% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 82.8% | — |
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. | — | 95.9% |
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. | 39% | — |
MMMU-Pro · with toolsMultimodal reasoning — multimodal understanding and reasoning with access to tools | 69.5% | — |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | 66.1% | — |
GraphWalks BFS (<128K)Long context — breadth-first traversal accuracy for inputs shorter than 128K tokens | 73.4% | — |
GraphWalks parents (<128K)Long context — parent-node retrieval accuracy for inputs shorter than 128K tokens | 50.8% | — |
| Overview | ||
| Company | OpenAI | Qwen |
| Release date | Mar 17 2026 | Sep 19 2024 |
| Access | Closed | Open Weight |
| Model details | View model | View model |
Other comparisons
GPT-5.4 nanovsClaude Fable 5.2Qwen2.5-MathvsClaude Fable 5.2GPT-5.4 nanovsGemini 4 NeonQwen2.5-MathvsGemini 4 NeonGPT-5.4 nanovsMuse EmberQwen2.5-MathvsMuse EmberGPT-5.4 nanovsGrok 4.7Qwen2.5-MathvsGrok 4.7GPT-5.4 nanovsDeepSeek-V4.1-Flash-1017Qwen2.5-MathvsDeepSeek-V4.1-Flash-1017GPT-5.4 nanovsMistral Large 4Qwen2.5-MathvsMistral Large 4Frequently asked questions
GPT-5.4 nano and Qwen2.5-Math don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. Only GPT-5.4 nano has a verified first-party price for an available API: $0.20 per million input tokens and $1.25 per million output tokens. No currently available pay-as-you-go API rate is tracked for Qwen2.5-Math. Qwen2.5-Math shipped 544 days before GPT-5.4 nano, so benchmark comparisons should account for the intervening progress.
Context windows are 400k (GPT-5.4 nano) vs 4k (Qwen2.5-Math). GPT-5.4 nano is closed, while Qwen2.5-Math is open weight.
Direct benchmark comparisons are unavailable — GPT-5.4 nano and Qwen2.5-Math don't publish scores on any of the same benchmarks.