Alibaba launches Qwen3.6 for agentic coding
In April 2026 Alibaba launched Qwen3.6, starting with the proprietary Qwen3.6-Plus for agentic coding on 2 April, followed by open Qwen3.6-35B-A3B weights in mid-April.
Total 671 · Clear filters
Microsoft AI releases MAI-Transcribe-1, MAI-Voice-1 and MAI-Image-2 in Foundry
On 2 April 2026 Mustafa Suleyman announced three Microsoft AI models for developers: MAI-Transcribe-1, MAI-Voice-1 and MAI-Image-2. All three became available in Microsoft Foundry and the MAI Playground, with published model cards.
Moonshot AI releases Kimi K2.6 for long-horizon coding
In April 2026 Moonshot AI released Kimi K2.6, an open-weight model aimed at long-horizon coding, with Agent Swarm scaled to 300 sub-agents and 4,000 coordinated steps.
Mistral AI releases Voxtral TTS, its first text-to-speech model
On 23 March 2026 Mistral AI released Voxtral TTS, a 4B text-to-speech model with voice cloning from three seconds of audio, open weights under a non-commercial licence, and API pricing of $0.016 per 1,000 characters.
Midjourney releases the V8 alpha
Midjourney released the V8 alpha on 17 March 2026, generating images about five times faster, with a native 2K --hd mode, improved text rendering, new prompt parameters and a Grid Mode interface. Point releases followed later that year.
Mistral AI releases Mistral Small 4
On 16 March 2026, Mistral AI released Mistral Small 4, a 119B mixture-of-experts model under Apache 2.0 that combines reasoning, image input and agentic coding in one model.
PixVerse closes a Series C and reaches unicorn status
PixVerse closed a Series C led by CDH Investments on 12 March 2026, saying the round made it a global AI unicorn and was Asia's largest in AI video generation. It reported more than 100 million users.
Reka releases the 7B edge vision-language model Reka Edge
On 11 March 2026 Reka released Reka Edge, a 7-billion-parameter vision-language model built for physical AI on devices rather than in the cloud, with a 3.5-bit quantization format and comparisons against Qwen3.5 9B, Cosmos Reason2 8B and Gemini 3 Pro.
Tencent launches the WorkBuddy desktop agent
Tencent launched WorkBuddy, a desktop AI agent for workplace tasks, on 9 March 2026. A global version followed in late May 2026.
OpenAI releases GPT-5.4 with built-in computer use
On 5 March 2026 OpenAI released GPT-5.4 in Thinking and Pro versions, its first general-purpose model with native computer use; it scored 75% on OSWorld-Verified, above the 72.4% human average.
Ai2 introduces Olmo Hybrid
On 5 March 2026 Ai2 released Olmo Hybrid, a fully open 7B model that stacks Gated DeltaNet linear-RNN layers with attention layers in a 3:1 ratio and reaches Olmo 3's MMLU accuracy with 49% fewer training tokens.
Microsoft releases Phi-4-reasoning-vision-15B
On 4 March 2026 Microsoft released Phi-4-reasoning-vision-15B, an MIT-licensed open-weight model that combines a reasoning backbone with a vision encoder and chooses how to answer each task.
MiniMax releases self-improving MiniMax-M2.7
In March 2026 MiniMax released MiniMax-M2.7, a model with 229B listed parameters that the company described as its first to take part in its own evolution, and which posted strong scores on machine-learning, coding, science and competition-math benchmarks.
Toyota Motor Manufacturing Canada contracts seven Digit robots
In February 2026, Toyota Motor Manufacturing Canada contracted seven Agility Robotics Digit robots under a robots-as-a-service deal, after a year-long pilot, to unload totes of auto parts in a plant building RAV4 SUVs.
Google releases Gemini 3.1 Pro
On 19 February 2026 Google released Gemini 3.1 Pro in preview, a model that scored 77.1% on the ARC-AGI-2 reasoning benchmark — more than double Gemini 3 Pro — and was made available the same day in the Gemini app, the Gemini API and Vertex AI.
Anthropic releases Claude Sonnet 4.6
On 17 February 2026 Anthropic released Claude Sonnet 4.6 at $3 and $15 per million tokens. It became the default model for Free and Pro users on claude.ai and Claude Cowork.
Alibaba releases natively multimodal Qwen3.5
In February 2026 Alibaba released Qwen3.5, a natively multimodal model family whose open flagship has 397B parameters and was posted to Hugging Face on 16 February under Apache 2.0.
ByteDance releases Doubao-Seed-2.0
On 14 February 2026 ByteDance released the Doubao-Seed-2.0 series in the Doubao app and on Volcano Engine, with Pro, Lite and Mini general agent models. Seed2.0 Pro was positioned against GPT-5.2 and Gemini 3 Pro.
MiniMax releases MiniMax-M2.5 and M2.5-Lightning
In February 2026 MiniMax released MiniMax-M2.5 together with M2.5-Lightning, a variant that runs at 100 tokens per second. MiniMax said the Lightning version costs $0.3 per million input tokens and $2.4 per million output tokens.
Zhipu releases GLM-5, a 744B open model
On 12 February 2026 Zhipu released GLM-5, a 744-billion-parameter open model under the MIT License. It uses DeepSeek Sparse Attention, and its pre-training data grew from 23 trillion to 28.5 trillion tokens.
ByteDance releases Seedance 2.0, prompting a Hollywood copyright backlash
ByteDance's Seedance 2.0 video model went live on 12 February 2026. Viral clips drew copyright complaints from the Motion Picture Association and Hollywood studios, and ByteDance later paused the global launch, according to The Information.