Compare models

Pick up to 4models. Specs render side-by-side. Share the URL — it's stateless.

Comparing agents instead? Switch to agent compare →

Selected

DeepSeek-V4-Pro×

Mistral Large 2×

Qwen3.6-27B×

DeepSeek-V3.1×Clear all

	DeepSeek-V4-Pro DeepSeek	Mistral Large 2 Mistral	Qwen3.6-27B Alibaba	DeepSeek-V3.1 DeepSeek
Vendor	DeepSeek	Mistral	Alibaba	DeepSeek
Family	DeepSeek	Mistral	Qwen	DeepSeek
Release date	2026-04-22	2024-07-24	2026-04-22	2025-08-21
Context window	1,000,000 tokens	128,000 tokens	262,144 tokens	128,000 tokens
Parameters	1.6T (49B active)	123B	27B (dense)	671B
Modality	text	text	text, vision, video	text
License	MIT	Mistral Research License	Apache-2.0	MIT
Source	open weights	open weights	open weights	open weights
Description	DeepSeek's flagship open-weight MoE. 1.6T parameters with 49B activated, 1M-token context, and a hybrid attention scheme (CSA + HCA) that delivers long-context inference at ~27% of V3.2's FLOPs.	Large text-only Mistral model with a 128K context window and 123B parameters, tuned for strong instruction following and long-context reasoning. Mistral's flagship open-weights release in the Large 2 line.	Alibaba's first dense open-weight in the 3.6 family. Strong agentic-coding scores (77.2 SWE-bench Verified, matching Claude 4.5 Opus on Terminal-Bench 2.0). Supports 201 languages and multimodal text/image/video input.	Large MoE open-weight model. Predecessor to DeepSeek-V4.
Links	weights →	vendor →weights →	weights →	weights →
Benchmarks
MMLU-Pro	84.2%	—	78.9%	—
GPQA-D	82.4%	—	—	—
HumanEval	95.1%	89.0%	91.5%	—
Aider	80.1%	—	70.2%	—
AIME-25	88.6%	—	—	—
LiveCB	72.4%	—	—	—
MMMU	—	—	70.3%	—