gpt.buzz

All models

20 tracked models across 8 vendors.

Frontier · 11

Open source · 9

DeepSeek-V4-Pro

DeepSeek

released 2026-04-22

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.

Context
1,000,000
Params
1.6T (49B active)
License
MIT
Source
open

Qwen3.6-27B

Alibaba

released 2026-04-22

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.

Context
262,144
Params
27B (dense)
License
Apache-2.0
Source
open

DeepSeek-V4-Flash

DeepSeek

released 2026-04-22

Smaller, faster sibling to DeepSeek-V4-Pro. Same 1M context window with a much lighter 284B / 13B-active MoE.

Context
1,000,000
Params
284B (13B active)
License
MIT
Source
open

DeepSeek-V3.1

DeepSeek

released 2025-08-21

Large MoE open-weight model. Predecessor to DeepSeek-V4.

Context
128,000
Params
671B
License
MIT
Source
open

Qwen3 235B

Alibaba

released 2025-04-29

Predecessor to the Qwen3.6 family.

Context
128,000
Params
235B
License
Apache-2.0
Source
open

Llama 4 Maverick

Meta

released 2025-04-05

Mixture-of-experts open-weight model from Meta.

Context
1,000,000
Params
400B
License
Llama 4 Community License
Source
open

Llama 4 Scout

Meta

released 2025-04-05

Context
10,000,000
Params
109B
License
Llama 4 Community License
Source
open

DeepSeek-R1

DeepSeek

released 2025-01-20

Reasoning-focused open-weight model.

Context
128,000
Params
671B
License
MIT
Source
open

Mistral Large 2

Mistral

released 2024-07-24

Context
128,000
Params
123B
License
Mistral Research License
Source
open