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-V3.1×

DeepSeek-R1×

Gemini 3.5×

Mistral Large 2×Clear all

	DeepSeek-V3.1 DeepSeek	DeepSeek-R1 DeepSeek	Gemini 3.5 Google	Mistral Large 2 Mistral
Vendor	DeepSeek	DeepSeek	Google	Mistral
Family	DeepSeek	DeepSeek	Gemini	Mistral
Release date	2025-08-21	2025-01-20	2026-05-20	2024-07-24
Context window	128,000 tokens	128,000 tokens	2,000,000 tokens	128,000 tokens
Parameters	671B	671B	—	123B
Modality	text	text	text, vision, audio, video	text
License	MIT	MIT	proprietary	Mistral Research License
Source	open weights	open weights	proprietary	open weights
Description	Large MoE open-weight model. Predecessor to DeepSeek-V4.	Reasoning-focused open-weight model.	Google DeepMind's next-gen Gemini — positioned as "frontier intelligence with action". Built for complex agentic workflows. Announced at Google I/O 2026.	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.
Links	weights →	weights →	vendor →	vendor →weights →
Benchmarks
MMLU-Pro	—	—	88.9%	—
GPQA-D	—	—	91.2%	—
HumanEval	—	—	—	89.0%
AIME-25	—	79.8%	96.1%	—
MMMU	—	—	86.0%	—