MAI Code 2 in Copilot: cheap tokens, close scores, one weak spot
Microsoft's MAI Code 2 models are in GitHub Copilot at half price until May 2027, and their benchmark averages sit close to GPT-6.1 Sol and Opus 5.5.
Read articleInsights
New AI models and major version releases: what changed, benchmarks with their caveats, pricing, access and limits.
Microsoft's MAI Code 2 models are in GitHub Copilot at half price until May 2027, and their benchmark averages sit close to GPT-6.1 Sol and Opus 5.5.
Read articleMicrosoft's decision model now reads images and gives steadier probabilities, and an early outside test backs most of its claims.
Read articleGemini 4 Neon costs US$3/US$15 per million tokens and benchmarks near the top, but it's slow, wordy and ignores sampling settings.
Read articleTencent's new open-weight Hy4 Flash uses about 12B of its 220B parameters per token and lists at US$0.07 per million input tokens.
Read articleOpenAI's GPT-6.2 Astra is 20% cheaper than GPT-6 Astra and Sol keeps its price, but long prompts, Ultrafast and weak legal scores need planning.
Read articleAlibaba's closed Qwen4-Max is live on QwenCloud with a 1M-token window and US$1.50/US$4.50 pricing, but most benchmarks are self-reported.
Read articleDeepMind's AlphaGenome 2 predicts DNA variant effects one cell type at a time, free for non-commercial research via API, with no weights yet.
Read articleMiniMax's H3-Image makes native 4K images from about $0.03 each via API, with open weights under a licence that excludes the US, EU and UK.
Read articleGemini 4 Hydrogen replaces 3.8 Flash at US$1.50/US$7.50 per million tokens, scores near frontier models and drops sampling controls.
Read articleElevenLabs' Scribe v3 models claim 27% fewer transcription errors at unchanged prices, with realtime speaker labels and US-default data processing.
Read articleMeta's Muse Ember costs $0.10 per million input tokens for agent subtasks, but whole requests over 200K tokens cost three times as much.
Read articleGoogle's Gemini 4 Carbon is now on the paid API, with stricter parameters, segmented long outputs and prices that double on 1 December.
Read articleAnthropic's Fable 5.2 cuts per-token prices by a fifth and lifts agentic scores, but four API changes can break code written for Fable 5.1.
Read articleThinking Machines' Inkling 2 Small scores well on coding and vision, but for now it runs only on Tinker, with a shorter context there and no weights yet.
Read article