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. | 284B | 27B/35B/122B |
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 | 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. | $0.0466Baidu | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $0.1399Baidu | — |
| $16.32Baidu · fp8 | — | |
| Benchmarks | ||
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. | 79% | 72% |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | 73.2% | 63.8% |
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. | 51.5% | 47.5% |
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. | 39% | — |
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. | — | 78.9% |
Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | 82.7% | — |
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. | — | 49.4% |
Toolathlon-VerifiedPersonal tool use — Tests how well the AI uses everyday personal tools and apps to get things done — a human-checked version of Toolathlon. Higher is better. | 70.3% | — |
IFEvalAgentic tasks — following verifiable instructions in generated responses | — | 93.4% |
CyberGymCybersecurity — Tests the AI on cybersecurity challenges — finding and exploiting software weaknesses inside a safe sandbox. Higher is better. | 76.7% | — |
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. | — | 25.3% |
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. | 61.4% | — |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | — | 86.6% |
IFBenchInstruction following — Tests whether the AI can follow detailed instructions and satisfy multiple constraints at once. Higher is better. | — | 76.1% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | — | 86.7% |
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. | — | 58% |
AutomationBenchBusiness workflows — Tests whether the AI can run real multi-step business workflows — the kind of end-to-end office processes companies want to automate — from start to finish. Higher is better. | 25.1% | — |
CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better. | — | 77.2% |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | — | 76.9% |
MMMUMultimodal — Tests the AI on understanding images and text together across many college subjects. Higher is better. | — | 83.9% |
| Overview | ||
| Company | DeepSeek | Qwen |
| Release date | Jul 31 2026 | Feb 24 2026 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek-V4-Flash-0731vsClaude Fable 5.2Qwen3.5 MediumvsClaude Fable 5.2DeepSeek-V4-Flash-0731vsGPT-6.2 AstraQwen3.5 MediumvsGPT-6.2 AstraDeepSeek-V4-Flash-0731vsGemini 4 NeonQwen3.5 MediumvsGemini 4 NeonDeepSeek-V4-Flash-0731vsMuse EmberQwen3.5 MediumvsMuse EmberDeepSeek-V4-Flash-0731vsGrok 4.7Qwen3.5 MediumvsGrok 4.7DeepSeek-V4-Flash-0731vsMistral Large 4Qwen3.5 MediumvsMistral Large 4Frequently asked questions
DeepSeek-V4-Flash-0731 leads Qwen3.5 Medium on 3 of the 3 benchmarks they both report (SWE-Bench Verified, BrowseComp, Humanity's Last Exam). Qwen3.5 Medium shipped 157 days before DeepSeek-V4-Flash-0731, so benchmark comparisons should account for the intervening progress.
DeepSeek-V4-Flash-0731 has 284B parameters, while Qwen3.5 Medium has 27B/35B/122B. Context windows are 1M (DeepSeek-V4-Flash-0731) vs 256k (Qwen3.5 Medium).
On SWE-Bench Verified, DeepSeek-V4-Flash-0731 leads at 79% vs Qwen3.5 Medium at 72%. On BrowseComp, DeepSeek-V4-Flash-0731 leads at 73.2% vs Qwen3.5 Medium at 63.8%. On Humanity's Last Exam · with tools, DeepSeek-V4-Flash-0731 leads at 51.5% vs Qwen3.5 Medium at 47.5%.