AIBase
EN

AI News

View More

NVIDIA Releases New Rubin Architecture: Super Chip Vera Rubin Appears, AI Computing Costs Will Drop Significantly

14.1k 12 hours ago
NVIDIA Releases New Rubin Architecture: Super Chip Vera Rubin Appears, AI Computing Costs Will Drop Significantly

Kechang Xunfei's Large Model Training Speeds Up Again! Domestic Computing Power Has Reached A100 Level!

10.3k 12 hours ago
Kechang Xunfei's Large Model Training Speeds Up Again! Domestic Computing Power Has Reached A100 Level!

OpenAI and Broadcom Collaborate on 10GW Custom Chips, Launch New AI Infrastructure in 2026

13.1k 10 hours ago
OpenAI and Broadcom Collaborate on 10GW Custom Chips, Launch New AI Infrastructure in 2026

The Domestic Computing Power Group Unites: The 10,000-Card Inference Engine and OpenSource Model Announced on the Same Day

9.7k 12 hours ago
The Domestic Computing Power Group Unites: The 10,000-Card Inference Engine and OpenSource Model Announced on the Same Day

AI Products

View More

MTIA

Meta Training and Inference Accelerator (MTIA) is Meta's self-developed next-generation AI training and inference acceleration chip. Compared to the previous generation, MTIA v2 has significantly improved computing power and memory bandwidth, enabling more efficient support for Meta's sorting and recommendation model applications. This is a significant achievement in Meta's continuous investment in self-developed hardware infrastructure, which will bring new capabilities to Meta's AI products and services.

Hardware infrastructure
6.6k

Models

View More

Grok 4 Fast

Xai

Grok 4 Fast

$1.4

Input tokens/M

$3.5

Output tokens/M

2k

Context Length

o3-mini

Openai

o3-mini

$7.7

Input tokens/M

$30.8

Output tokens/M

200

Context Length

Claude Haiku 4.5

Anthropic

Claude Haiku 4.5

$7

Input tokens/M

$35

Output tokens/M

200

Context Length

Gemini 2.5 Flash

Google

Gemini 2.5 Flash

$2.1

Input tokens/M

$17.5

Output tokens/M

1k

Context Length

Gemini 2.5 Flash-Lite

Google

Gemini 2.5 Flash-Lite

$0.7

Input tokens/M

$2.8

Output tokens/M

1k

Context Length

Qianfan-Lightning

Baidu

Qianfan-Lightning

-

Input tokens/M

-

Output tokens/M

128

Context Length

Qwen3-Next-80B-A3B-Instruct

Alibaba

Qwen3-Next-80B-A3B-Instruct

$2

Input tokens/M

-

Output tokens/M

256

Context Length

Doubao-1.5-pro-32k

Bytedance

Doubao-1.5-pro-32k

$0.8

Input tokens/M

$2

Output tokens/M

128

Context Length

Hunyuan-T1-20250822

Tencent

Hunyuan-T1-20250822

$1

Input tokens/M

$4

Output tokens/M

32

Context Length

Hunyuan-T1-latest

Tencent

Hunyuan-T1-latest

$1

Input tokens/M

$4

Output tokens/M

32

Context Length

DeepSeek-V3.1

Deepseek

DeepSeek-V3.1

$4

Input tokens/M

$12

Output tokens/M

128

Context Length

Qwen3-1.7B

Alibaba

Qwen3-1.7B

-

Input tokens/M

-

Output tokens/M

32

Context Length

gpt-oss-20b

Openai

gpt-oss-20b

$0.4

Input tokens/M

-

Output tokens/M

128

Context Length

GPT-5 mini

Openai

GPT-5 mini

$1.75

Input tokens/M

$14

Output tokens/M

400

Context Length

GPT-5 nano

Openai

GPT-5 nano

$0.35

Input tokens/M

$2.8

Output tokens/M

400

Context Length

Qwen3-235B-A22B-Instruct-2507

Alibaba

Qwen3-235B-A22B-Instruct-2507

$2

Input tokens/M

-

Output tokens/M

64

Context Length

Pangu-NLP-N2-128K-5.0.1.1

Huawei

Pangu-NLP-N2-128K-5.0.1.1

-

Input tokens/M

-

Output tokens/M

128

Context Length

Pangu-NLP-N2-32K-5.0.1.1

Huawei

Pangu-NLP-N2-32K-5.0.1.1

-

Input tokens/M

-

Output tokens/M

32

Context Length

GLM-4.5-Flash

Chatglm

GLM-4.5-Flash

-

Input tokens/M

-

Output tokens/M

128

Context Length

GLM-4.5-AirX

Chatglm

GLM-4.5-AirX

-

Input tokens/M

-

Output tokens/M

128

Context Length

Kimi K2 Thinking MLX 4.25bit

inferencerlabs

K

A text generation model implemented based on the MLX library, supporting inference in multiple quantization methods, with distributed computing capabilities, and can run efficiently in the Apple hardware environment.

Natural Language ProcessingMlxMlxEnglish
inferencerlabs
1.5k
2

DeepSeek R1 0528 Qwen3 8B GGUF

Sci-fi-vy

D

DeepSeek-R1-0528 is a minor version upgrade model of the DeepSeek R1 series. It significantly improves the inference depth and ability through increased computing resources and algorithm optimization, and performs excellently in multiple benchmark tests such as mathematics and programming.

Natural Language ProcessingTransformersTransformers
Sci-fi-vy
1.2k
1

Llama 3 8b Quantized

SweatyCrayfish

L

The 4-bit quantized version of the Llama 3 model, which optimizes memory usage and speeds up inference, suitable for environments with limited computing resources.

Natural Language ProcessingTransformersTransformersEnglish
SweatyCrayfish
2k
11
AIBase
Empowering the future, your artificial intelligence solution think tank
English简体中文繁體中文にほんご
FirendLinks:
AI Newsletters AI ToolsMCP ServersAI NewsAIBaseLLM LeaderboardAI Ranking
© 2026AIBase
Business CooperationSite Map