https://attapall.ai Aggregated change feed for AI model releases, pricing updates, deprecations, and verified-source notes across 12 vendors. en-US Mon, 11 May 2026 17:10:06 +0000 https://attapall.ai https://attapall.ai/#changes/2026-05-10-shell-1 Sun, 10 May 2026 00:00:00 +0000 infra Volatile facts (model IDs, prices, dates, status) extracted to data/models.json. Manuals render lineup tables from JSON. Weekly freshness sweep will append entries here automatically. https://docs.x.ai/docs/release-notes https://attapall.ai/#changes/2026-04-30-xai-1 Thu, 30 Apr 2026 00:00:00 +0000 new_model Released 2026-04-30. $1.25/$2.50 — cheapest hosted frontier on list price. 1M context, native video input, always-on reasoning. ~40% input price cut vs Grok 4. https://api-docs.deepseek.com/news/news260424 https://attapall.ai/#changes/2026-04-24-deepseek-1 Fri, 24 Apr 2026 00:00:00 +0000 new_model V4 Pro: 1.6T / 49B-active MoE, 1M context, MIT-licensed. Open-source SOTA on agentic-coding benchmarks. V4 Flash: 284B / 13B-active, same context. 75% Pro discount runs through 2026-05-31. https://help.openai.com/en/articles/9624314-model-release-notes https://attapall.ai/#changes/2026-04-23-openai-1 Thu, 23 Apr 2026 00:00:00 +0000 new_model Available in ChatGPT (Plus/Pro/Business/Enterprise) + Codex now. Built around long-running goal completion. API pricing TBA. https://huggingface.co/moonshotai https://attapall.ai/#changes/2026-04-20-moonshot-1 Mon, 20 Apr 2026 00:00:00 +0000 new_model Modified MIT license. 1T total / 32B active. 262K context, 16K max output. Agent Swarm: 300 sub-agents × 4000 steps per run. $0.60/$2.50 on Moonshot API. https://www.anthropic.com/news https://attapall.ai/#changes/2026-04-16-claude-1 Thu, 16 Apr 2026 00:00:00 +0000 new_model $5/$25 per M tokens. 200K context. New xhigh effort dial. Major vision lift (98.5% visual-acuity). Task budgets (public beta), file-system memory. Tokenizer change: 1.0–1.35× more tokens than 4.6. https://huggingface.co/zai-org https://attapall.ai/#changes/2026-04-08-zai-1 Wed, 08 Apr 2026 00:00:00 +0000 new_model 745B / 44B-active MoE, MIT license, 200K context. DeepSeek Sparse Attention integrated. From China's first publicly-traded AI lab. Subscription preview late March. https://modelstudio.console.alibabacloud.com https://attapall.ai/#changes/2026-04-alibaba-1 Wed, 15 Apr 2026 00:00:00 +0000 new_model Qwen 3.6 Max (proprietary flagship, 1M+ context, agentic + visual reasoning). Qwen 3.6 Plus (high-perf, speed/cost tier). Wan 2.7 (video gen) + HappyHorse 1.0 (image-to-video, top-ranked) added to Model Studio. https://docs.x.ai/docs/release-notes https://attapall.ai/#changes/2026-03-31-xai-2 Tue, 31 Mar 2026 00:00:00 +0000 new_model $2.00/$6.00 per M tokens. 2M context — largest in the field alongside Gemini 3.1 Pro. Strong agentic tool calling. https://help.openai.com/en/articles/9624314-model-release-notes https://attapall.ai/#changes/2026-03-05-openai-1 Thu, 05 Mar 2026 00:00:00 +0000 new_model 75.0% on OSWorld-Verified (vs 47.3% for GPT-5.2). Beats human baseline 72.4%. 95% first-try / 100% within 3 tries on real portals. https://huggingface.co/Lightricks/LTX-2.3 https://attapall.ai/#changes/2026-03-05-lightricks-1 Thu, 05 Mar 2026 00:00:00 +0000 new_model 22B Diffusion Transformer. Native 4K @ 50fps with synchronized audio. FP8 quantized fits 24GB VRAM (RTX 4090/5090). ~$0.04/sec hosted. https://ai.google.dev/gemini-api/docs/changelog https://attapall.ai/#changes/2026-02-19-google-1 Thu, 19 Feb 2026 00:00:00 +0000 new_model $2/$12 per M (≤200K input); $4/$18 above. 77.1% on ARC-AGI-2 (vs 31.1% for Gemini 3 Pro). 94.3% on GPQA Diamond. Deep Think mode. https://seed.bytedance.com/en https://attapall.ai/#changes/2026-02-12-bytedance-1 Thu, 12 Feb 2026 00:00:00 +0000 new_model Accepts up to 9 image / 3 video / 3 audio references per prompt. 4-15s multi-shot output with dual-channel audio. Most flexible reference inputs in video gen. https://huggingface.co/zai-org https://attapall.ai/#changes/2026-02-11-zai-1 Wed, 11 Feb 2026 00:00:00 +0000 new_model 745B total / 40-44B active MoE. 200K context. Pre-training data 28.5T tokens (up from 23T on GLM-4.5). https://huggingface.co/Qwen https://attapall.ai/#changes/2026-02-alibaba-1 Sun, 08 Feb 2026 00:00:00 +0000 price_change Open-source + API. Native multimodal (text + image + video). Includes Qwen 3.5 Omni for full audio + multimodal.