Skip to main content
AI Choice EngineAI Choice Engine
Browse decision tools

Source receipts

See what each claim is based on

The catalog separates checked comparisons from release signals. These dated or explicitly qualified source links are the receipts behind launch notes, open-weight coverage, and the watchlist.

This page is a receipt index for editorial and release claims, not a per-field audit of every model, hardware, or price value. Those catalog records show their own check dates and should be confirmed with the linked provider or retailer before purchase.

Launch guides

Release-date receipts for current model and coding-assistant coverage, with source type shown where classified.

  • GPT-5.6 Sol, Terra, and Luna: what the three-tier release changes

    Released July 9, 2026 · OpenAI · source type not classified

    OpenAI model release notes

  • Claude Opus 5: where the new frontier tier earns its price

    Released July 24, 2026 · Anthropic · source type not classified

    Anthropic release notes

  • Gemini 3.6 Flash: the efficient multimodal workhorse

    Released July 21, 2026 · Google · source type not classified

    Google model announcement

  • Gemini 3.7 Flash: the new coding and agent workhorse

    Released August 13, 2026 · Google · source type not classified

    Google Gemini 3.7 Flash announcement

  • Kimi K3: open weights, huge context, and a datacentre-sized trade-off

    Released July 16, 2026 · Moonshot AI · source type not classified

    Kimi K3 announcement

  • DeepSeek V4-Flash-0731: why the post-training update matters

    Released July 31, 2026 · DeepSeek · secondary

    DeepSeek V4-Flash release analysis

  • Qwen3.8-Max: hosted API versus the published open-weight checkpoint

    Released August 3, 2026 · Qwen · official

    Alibaba Cloud official release

  • Qwen3.8-27B: Apache 2.0 dense VLM for local work

    Released August 14, 2026 · Qwen · source type not classified

    Qwen3.8-27B on Hugging Face

  • Muse Spark 1.2: a model co-trained with its coding harness

    Released August 5, 2026 · Meta · source type not classified

    Meta AI Research announcement

  • Muse Code: Meta's beta terminal agent with persistent background workers

    Released August 5, 2026 · Meta · source type not classified

    Meta Muse Code product page

  • Gemini 3.1 Pro: long context and multimodal input in limited preview

    Released February 19, 2026 · Google · source type not classified

    Google AI model documentation

  • Grok 4.5: mid-tier pricing with real-time data hooks

    Released July 16, 2026 · SpaceXAI · source type not classified

    SpaceXAI Grok documentation

  • Claude Sonnet 5: mid-tier tokens that can hide agentic cost

    Released June 30, 2026 · Anthropic · source type not classified

    Anthropic release notes

  • Claude Haiku 4.5: volume tier with a smaller context envelope

    Released October 15, 2025 · Anthropic · source type not classified

    Anthropic model overview

  • GLM 5.2: open weights with a million-token context

    Released June 16, 2026 · Z.ai · source type not classified

    Z.ai GLM documentation

  • GLM 5.3: Coding Plan flagship, API and weights still pending

    Released August 14, 2026 · Z.ai · source type not classified

    Z.ai GLM-5.3 announcement

  • Kimi K2.6: open-weight volume with a smaller context window

    Released April 21, 2026 · Moonshot AI · source type not classified

    Moonshot AI Kimi platform

Open-weight coverage

Source links for licences, checkpoints, and deployment notes; provider and secondary receipts are not interchangeable.

Release watchlist

Interesting release signals that are not yet included as verified catalog comparisons.

  • Qwen3.8-Max / Qwen3.8-2.4T-A95B / Qwen3.8-27B

    Qwen · Hosted API plus two Hugging Face checkpoints are published — keep them as separate identities

    Qwen3.8-2.4T-A95B on Hugging Face

  • Amazon Nova Premier / Canvas / Reel (Bedrock)

    Amazon · Legacy — EOL 2026-09-14 (Premier) / 2026-09-30 (Canvas/Reel)

    AWS Bedrock model lifecycle

  • Cohere Command R / Command R+ (Bedrock)

    Cohere · EOL on Bedrock 2026-08-19

    AWS Bedrock model lifecycle

  • EU AI Act Article 50 transparency

    EU / UK buyers · Live since 2026-08-02; watermark grace to 2026-12-02; high-risk deferred under Omnibus

    EU AI transparency guidelines

  • Relay.app shutdown

    Relay · EOL — free wipe 2026-08-15 23:59 PT; paid 2026-09-14

    Relay.app shutdown notice

  • Gemini 2.5 Pro / Flash retirement

    Google · Replacement path to Gemini 3.x; EOL mid-Oct 2026 (Google docs conflict Oct 16 vs Oct 20)

    Google Cloud model versions & lifecycle

  • xAI Imagine Image 2.0

    xAI · Quality Mode GA; grok-imagine-image-2.0 is $0.04/image on the xAI API

    xAI models docs

  • v0 (Vercel) plan ladder

    Vercel · Free / Premium $20 sunsetting / Plus ~$30 / Business ~$100

    v0 pricing docs

  • Gemini 3.5 Flash Cyber

    Google · Limited or preview access

    Google model announcement

  • OpenAI Daybreak / GPT-5.6-Cyber

    OpenAI · Defender-focused; HSK path from Sep 2026

    OpenAI Daybreak

Benchmark sources

Original leaderboards, independent evaluation pages, and the dated local snapshots rendered in the benchmark library.

Looking for the latest dated list? Browse new model releases.