New AI Models

DeepSeek-V4-Flash-0731

DeepSeek-V4-Flash-0731 is the official public beta of the DeepSeek-V4-Flash API, released on July 31, 2026 under the model name deepseek-v4-flash. It delivers significantly stronger agent performance, scoring well above V4-Pro-Preview on benchmarks such as Terminal Bench 2.1 (82.7), DeepSWE, and Cybergym. The model keeps the same architecture and size as V4-Flash-Preview, with only re-post-training applied, and it natively supports the Responses API format with specific adaptation for Codex.

HY-ASR-3-NoStream

Hunyuan HY-ASR-3-NoStream is the non-streaming ASR model from Tencent Hunyuan, the file transcription form of Hy ASR 3.0 preview. Built on the Hy3 foundation model, it combines high-precision recognition with deep semantic understanding and supports Mandarin, English and 20 Chinese dialects. It uses context to correct homophones and resolve ambiguity, stays robust in noisy and whispered speech, and suits batch transcription of meetings, interviews and media recordings.

HY-ASR-3-Stream

HY-ASR-3-Stream is Tencent Hunyuan's streaming ASR model, released in preview as Hy-ASR-3.0-preview and built on the Hy3 LLM base. It transcribes Mandarin, English, and 20 Chinese dialects in a single engine and returns results over WebSocket in real time, producing text as the speaker talks. Deep semantic understanding improves general recognition, context awareness, and dialect coverage, suiting low-latency scenarios such as instant messaging transcription, live call captions, and voice assistant command parsing.

LongCat-Flash-Lite-Sparse

LongCat-Flash-Lite-Sparse is a non-thinking Mixture-of-Experts LLM from Meituan with 69B total and roughly 3B activated parameters per token. Built on LongCat-Flash-Lite, it replaces dense MLA with LongCat Sparse Attention (LSA), a hardware-efficient sparse attention framework extending DeepSeek Sparse Attention, and natively supports context windows up to 1M tokens. It improves long-context inference efficiency and agentic coding, search, and tool-use performance over its dense predecessor, and can be served on a single node such as one H20-141G GPU, making it a fit for cost-effective long-context and agent workloads.

Gemini Robotics ER 2

Gemini Robotics ER 2 is Google DeepMind's most capable embodied reasoning model, a vision language model that acts as the robot's high-level brain. It observes the environment, plans multi-step tasks that last several minutes, and coordinates with VLA models to execute them, with support for multi-robot collaboration. Google describes it as its safest robotics model to date in safety constraint following and human proximity benchmarks. It is available in Google AI Studio and in private preview on the Gemini Enterprise Agent Platform.

MiniMax H3

MiniMax H3 is a general-purpose multimodal generation model that understands unified context across text, images, video, and audio, generating video with native stereo sound at up to 15 seconds and native 2K resolution. Built on components such as H3-VAE and the H3-Omni Transformer, it handles instruction following, accurate text and brand rendering, and V2V motion transfer. Per-second pricing at 2K is less than a third of mainstream models, and open weights are planned. It targets advertising, e-commerce, product design, UI/UX, and gaming.

P-Image-Ideogram

P-Image-Ideogram is Ideogram's fast text-to-image model family built with Pruna AI, tuned for a Pareto optimal balance of quality, latency, and cost. It exposes five Quality levels from VERY_LOW to VERY_HIGH: at 1K resolution it typically returns in about 2.9 to 5.0 seconds for roughly $0.003 to $0.015 per image, with a Very High level at $0.033 per 1K image for the toughest briefs. On the third-party Design Arena leaderboard its Very Low and Low 1K modes land on the quality-cost Pareto frontier, and the API offers quality, resolution (1K or 2K), and aspect_ratio controls for high-volume production and iteration.

Qwen-Audio-3.0-ASR-Flash

Qwen-Audio-3.0-ASR-Flash is a non-realtime speech recognition model from Alibaba's Qwen team, available on the Bailian platform for transcribing audio files over HTTP, with up to 5 minutes and 2GB per request. It supports 30 languages including Chinese, English, Japanese and Korean, covers the seven major Chinese dialect groups, and offers hotwords and context prompting to improve accuracy on specific terms. Output adds punctuation prediction and text normalization, with optimized classical poetry recognition for use cases such as call-center recordings, podcasts and audiobooks.

qwen-audio-3.0-asr-flash-filetrans

qwen-audio-3.0-asr-flash-filetrans is a non-real-time ASR model from Alibaba Cloud Bailian built for transcribing complete audio and video files, with asynchronous processing for recordings up to 12 hours and 2GB each. It integrates Context augmentation and strengthens recognition of industry-specific hotwords, covering 30 languages including Chinese, English, Japanese, and Korean. Typical uses include short-video subtitles, customer service quality inspection, and financial dual recording, billed at 0.00022 CNY per second of audio.

Qwen-Audio-3.0-ASR-Flash-Streaming

Qwen-Audio-3.0-ASR-Flash-Streaming is a flagship speech recognition model from Tongyi Lab, built for low-latency, high-concurrency real-time interaction and returning text while the speaker is still talking. It supports 30 languages including Chinese, English, Japanese and Korean, plus seven major Chinese dialect groups, and combines context enhancement, hotwords, and specialized training for industry vocabulary. Typical uses include real-time meeting interpretation, contact center agent assistance, and voice assistants.

Lyria 3.5

Lyria 3.5 is Google's latest AI music generation model, announced on July 29, 2026. It improves on musicality, lyrics, and vocals: richer and more natural melodic structures, higher-quality lyrics with better prompt adherence and structural awareness, and more realistic, emotionally nuanced vocals with clearer pronunciation. Users also get easier control over the tempo and duration of generated tracks. Lyria 3.5 is available in Google Flow Music.

Grok Voice Think Fast 2.0

Grok Voice Think Fast 2.0 is xAI's next-generation speech-to-speech voice model built for real-time voice agents. It reasons while speaking, hitting 0.70s time to first audio and topping Artificial Analysis' Speech-to-Speech Quality Index at 82.9%, ahead of GPT-Realtime-2.1 and Gemini 3.1 Flash. Transcription accuracy beats dedicated models like Deepgram Nova 3 by 1.5-2.0x, widening to roughly 10x in noisy settings. Available via the xAI API at $0.08 per minute of audio, it fits customer support, sales, and agentic telephony workloads.

Claude Opus 5

Claude Opus 5 is a text model from Anthropic that comes close to the frontier intelligence of Claude Fable 5 at half the price, and is now the default model on Claude Max. It leads all models on Frontier-Bench v0.1, more than doubling Opus 4.8's score at lower cost per task, and scores three times the next-best model on ARC-AGI 3. Built for long-running, multi-step agent work, it verifies its own output and iterates until tasks succeed, making it well suited to software engineering, research, and knowledge work.

V8.2

V8.2 is Midjourney's image model released in July 2026, with the update focused on aesthetics, image quality, and personalization. Compared with V8, it generates more creative, sophisticated results and sharply reduces the random low-quality images users occasionally saw before. Personalization also reads individual taste more accurately, especially for profiles with many ratings, and draws on a larger, improved image pool. It is now live on midjourney.com.

FLUX 3

FLUX 3 is Black Forest Labs' multimodal foundation model that learns from images, videos, and audio within a unified architecture, with all capabilities built on a single multimodal flow matching model. For image work it synthesizes and edits across a wide range of styles, aspect ratios, and resolutions, and renders high-accuracy text in multiple languages, with marked gains over earlier FLUX versions on complex prompts and text generation. Access rolls out through Early Access, followed by APIs and private weights, plus an open-weight FLUX 3 Dev backbone.

MAI-Image-2.5 Pro

MAI-Image-2.5 Pro is Microsoft's high-fidelity image generation model, introduced in July 2026 as the premium variant of MAI-Image-2.5 within the MAI model family. Positioned as Microsoft's highest-fidelity image model to date, it targets workflows where quality matters most, with precise in-image text rendering, natural-language editing, and support for hero imagery and detailed editing. It is available in public preview through the Azure AI Foundry model catalog and can be tried in the MAI Playground.

MAI-Voice-2-Flash

MAI-Voice-2-Flash is Microsoft's fast, efficient text-to-speech model built for speed and scale in high-volume voice experiences. It runs twice as fast as MAI-Voice-2 at 32% lower cost ($15 per 1M characters) while retaining the family's natural prosody and high acoustic quality. Available in public preview via the Azure AI Foundry model catalog and integrated into Azure Voice Live, it already powers Dynamics 365 Contact Center and suits latency-sensitive use cases like voice agents, call centers, and narration.

3.5 Flash Cyber

Gemini 3.5 Flash Cyber is Google's lightweight cybersecurity model, fine-tuned from Gemini 3.5 Flash to find, validate, and patch code vulnerabilities more cost-efficiently than large security models. It scores competitive pass@1 results against much larger models on the CyberGym benchmark and surpasses mainline 3.5 Flash and 3.6 Flash on Google's Big Sleep evaluation. Access is limited to governments and trusted partners through the CodeMender pilot.

3.5 Flash-Lite

Gemini 3.5 Flash-Lite is a low-latency, cost-effective multimodal model from Google, released as a stable version for production use on July 21, 2026. It accepts text, image, video, audio, and PDF inputs, with an input token limit of 1,048,576. Optimized for subagent tasks and document parsing, it suits high-volume agentic workflows and simple data extraction where latency and API cost are the primary constraints.

Gemini 3.6 Flash

Gemini 3.6 Flash is Google's fast, cost-efficient multimodal model, generally available as of July 21, 2026. It accepts text, image, video, audio, and PDF input within a 1M token context window. Compared with 3.5 Flash, it delivers higher token efficiency and stronger code and agent planning at a lower price, while addressing developer feedback on output verbosity. It excels at code generation, agentic execution, and spatial reasoning, making it a fit for rapid, iterative agent workflows.