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. | 671B | 123B |
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. | 128k | 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. | $0.27SiliconFlow | $0.40Mistral |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $1.00SiliconFlow | $2.00Mistral |
| $104SiliconFlow · fp8 | $180Mistral | |
| 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. | 68.4% | 72.2% |
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. | 57.8% | 61.3% |
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. | 74.9% | — |
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. | — | 32.6% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 80.7% | — |
| Overview | ||
| Company | DeepSeek | Mistral |
| Release date | Sep 22 2025 | Dec 9 2025 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek V3.1 TerminusvsClaude Haiku 5.5Devstral 2 (123B)vsClaude Haiku 5.5DeepSeek V3.1 TerminusvsGPT-6.1 SolDevstral 2 (123B)vsGPT-6.1 SolDeepSeek V3.1 TerminusvsGemini 4 ArgonDevstral 2 (123B)vsGemini 4 ArgonDeepSeek V3.1 TerminusvsMuse Spark OpenDevstral 2 (123B)vsMuse Spark OpenDeepSeek V3.1 TerminusvsGrok 4.7Devstral 2 (123B)vsGrok 4.7DeepSeek V3.1 TerminusvsKimi K3Devstral 2 (123B)vsKimi K3Frequently asked questions
Devstral 2 (123B) leads DeepSeek V3.1 Terminus on 2 of the 2 benchmarks they both report (SWE-Bench Verified, SWE-Bench Multilingual). DeepSeek V3.1 Terminus shipped 78 days before Devstral 2 (123B), so benchmark comparisons should account for the intervening progress.
DeepSeek V3.1 Terminus has 671B parameters, while Devstral 2 (123B) has 123B. Context windows are 128k (DeepSeek V3.1 Terminus) vs 256k (Devstral 2 (123B)).
On SWE-Bench Verified, Devstral 2 (123B) leads at 72.2% vs DeepSeek V3.1 Terminus at 68.4%. On SWE-Bench Multilingual, Devstral 2 (123B) leads at 61.3% vs DeepSeek V3.1 Terminus at 57.8%.